Child Abuse Victimization,Depression,and Substance Use Among Homeless Women: Application of General Strain Theory

Abstract

The main purpose of this study was to investigate the association between child abuse and substance use among homeless women based on a framework of General Strain Theory, which emphasizes the strong relationship between deviant behaviors and personally experienced strain. In this study, as the strain associated with substance use, child abuse experience in the past was assessed. This research tested three hypotheses about (a) the association between physical and sexual abuse during childhood and substance use (heavy drinking, marijuana use, and crack cocaine use), (b) the mediating effect of depressive symptoms, and (c) the moderating effect of positive social supports on the relationship between abuse during childhood and substance use with a survey and interview data of a randomly selected sample of 445 homeless women in a temporary shelter setting in Los Angeles County between June 2007 and March 2008. Although the results indicated that abuse during childhood was not significantly associated with any form of substance use, the results of multivariate analyses indicated that depressive symptoms fully mediated the four relationships (p < .01; physical abuse and heavy drinking, physical abuse and crack cocaine use, sexual abuse and heavy drinking, and sexual abuse and crack cocaine use). The results of this study also indicated that positive social supports significantly moderate the effect of physical (p < .01) and sexual (p < .05) abuse experience on heavy drinking, which implies that positive social supports would be significant to reduce substance use among homeless women who had a history of physical and sexual abuse during childhood.

Keywords

child abuse substance use homeless women general strain theory

The association between abuse during childhood and substance use during adulthood among women has been consistently examined (Herrenkohl, Hong, Klika, Herrenkohl, & Russo, 2013; Ross & Dennis, 2009). Child abuse may play a significant role in substance use among homeless women (Browne, 1993; Chen, Tyler, Whitbeck, & Hoyt, 2004; Stein, Leslie, & Nyamathi, 2002; Tyler & Melander, 2015). Agnew’s (1992) General Strain Theory (GST) provides a theoretical framework to understand how child abuse contributes to homeless women’s substance use. This particular theory focuses on how strain leads to deviant behaviors such as substance use and crime (Agnew, 1992, 2006, 2009). GST focuses on negative emotions as reactions to various noxious strains and highlights the mediating role of those emotions in the relationship between strain and deviant behaviors. In addition, GST outlines the likelihood that deviant behaviors may be enhanced or diminished by a range of conditioning factors that could be treated as moderators.

This study focused on physical and sexual child abuse as strains that may contribute to substance use among homeless women. Depression was examined as a negative emotion that may mediate the link between victimization and substance use in this population. Analyses examined whether positive social support, a conditioning factor, inhibits substance use as a response to victimization among homeless women. This theory-based examination of mediators and conditioning factors may lead to informed and effective efforts to address substance use in this population.

Theoretical Background

General Strain Theory

Deviant behaviors such as substance use, delinquency, and crime are predicted by strain (Agnew, 1992). The definition of objective strain is “events or conditions that are disliked by most members of a given group” (Agnew, 2001, p. 320) and the definition of subjective strain is “events or conditions that are disliked by the people who are experiencing (or have experienced) them” (Agnew, 2001, p. 321). GST emphasizes the strong relationship between deviant behaviors and personally experienced strain such as child abuse and neglect, verbal and physical abuse by a partner, homelessness, and criminal victimization (Agnew, 2001, 2009). The likelihood of engaging in deviant behavior increases based on the intensity and duration of strains (Agnew, 1992).

One of the strengths of GST is its argument that negative emotion plays a mediating role in predicting deviant behaviors. Strain increases negative emotional states, including depression, fear, anger, and frustration, which partly explain the effects of strain on crime (Agnew, 1992, 2001, 2006, 2009). Agnew (1992) emphasized the importance of examining nonstrain variables as conditioning factors, arguing that previous iterations of strain theory failed to explain why all individuals do not engage in deviant behaviors when experiencing the same strains. Conditioning factors have been examined as moderators that attenuate or increase the influence of strain on deviant behavior and alleviate or exacerbate the resulting negative emotions (Agnew, 2009; Agnew & White, 1992; Jang & Rhodes, 2012).

In sum, GST articulates two sets of processes, mediating and conditioning, through which deviant behaviors can be explained. This study examined the mediating effect of depression and moderating effect of positive social support on the relationship between violent victimization and substance use among homeless women.

Child Abuse Victimization and Substance Use

Abuse during childhood may cause short-term (instant) negative effects, but also may result in long-term effects that extend into adulthood (Browne, 1993) as a long-term life-threatening strain on a person’s life. Using the GST framework, child abuse victimization could be described as being treated in a negative manner by others. Child abuse victimization was identified as a specific strain, which is most likely to cause deviant behaviors (Agnew, 2001; Grothoff, Kempf-Leonard, & Mullins, 2014). Particularly, women have been described as more likely to value close relationships and to be more vulnerable to violence perpetrated by partners, parents, or close friends (Broidy & Agnew, 1997; Jones, Worthen, Sharp, & McLeod, 2018). Given this background, this study focused on physical and sexual abuse.

Child abuse may lead to substance use among homeless women. Nearly 70% of homeless women experienced physical abuse during childhood, and more than 40% reported sexual abuse during childhood (Bassuk et al., 1996; Green et al., 2012). Homeless women are more likely than housed women to have been victimized during childhood (Jasinski, Wesely, Mustaine, & Wright, 2005). The association between child abuse and substance use among homeless women has been empirically examined (Tyler & Melander, 2015), yet their link may be more complicated. Stein et al. (2002) found that childhood maltreatment is indirectly associated with drug use via the mediating effects of self-esteem and depression among homeless women. Chen et al. (2004) also found that early sexual abuse among homeless women and runaway adolescent girls indirectly predicts drug use. These results indicated that unlike the consistent association between child abuse and substance use among women in general, the association among homeless women remains unclear. GST provides a theoretical framework to understand this link by including negative emotion and social support as a potential mediator and moderator, respectively.

Mediating and Moderating Factors: Negative Emotion and Positive Social Support

GST emphasizes the mediating role of negative emotion on the relationship between strain and deviant behaviors. Individuals experience different emotions based on types of strain. Individuals who feel depressed are more likely to use alcohol and illicit drugs to alleviate the negative emotion (Agnew, 2006, 2009; Piquero & Sealock, 2004). Women are likely to respond to depression and strains with self-directed deviant behaviors, especially substance use, such as alcohol and illicit drug use (Jang, 2007; Kaufman, 2009; Manasse & Ganem, 2009; Moon & Morash, 2017). History of abuse during childhood, the specific strain to be examined in this study, increases the likelihood of depressive symptoms among women (Najdowski & Ullman, 2011; Shorey et al., 2011). Homeless women experience depressive symptoms associated with child abuse victimization (Stein et al., 2002), and depressive symptoms predict substance use among homeless women (Galaif, Nyamathi, & Stein, 1999; Upshur, Jenkins, Weinreb, Gelberg, & Orvek, 2017).

In addition, GST highlights that not all individuals respond to strain with deviant behaviors because of conditioning (i.e., moderating) factors that may protect against deviant behaviors (Agnew, 1992, 2006, 2009). Of particular interest in this study was positive social support as a conditioning factor. Agnew (1992) suggested that social support is an important conditioning factor that alleviates deviant behaviors in response to strain. Evidence indicated that individuals, particularly those with high levels of social support, may alleviate strain through their social support networks rather than by engaging in deviant behaviors (Kort-Butler, 2010). Those factors may condition the relationships between strain, negative emotionality, and deviant behaviors. Social support plays a key role in moderating the effects of childhood victimization on healthy and psychological status as a conditional factor (Folger & Wright, 2013). Homeless women are no exception; positive social support networks play a key role in reducing the likelihood of drug dependence (Galaif et al., 1999; Stein, Dixon, & Nyamathi, 2008).

Research Hypotheses

This study focused on child abuse as a strain that may lead to substance use among homeless women. Despite evidence indicating that child abuse increases substance use among women, little attention has been paid to the mediating role of negative emotionality and potential conditioning factors such as social support. Moreover, previous studies on the effects of child abuse and substance use among homeless women have produced inconsistent results. The association between child abuse and substance use among the general female population has been consistently supported in previous studies but the association among homeless women is quite unclear. This requires further study focusing on the complicated path to substance use rather than focusing on a direct path from child abuse victimization to substance use (Chen et al., 2004; Stein et al., 2002). This also suggests further investigation is needed to improve our understanding on this topic. Following the recommendations of previous studies (Agnew, 1992, 2006, 2009), this study examined the mediating effect of depression and the role of one conditioning factor proposed in GST, positive social support:

Hypothesis 1: The experience of physical and sexual child abuse victimization will increase the likelihood of women reporting current substance use.

Hypothesis 2: Negative emotion (depressive symptoms) will mediate the relationship between victimization and current substance use.

Hypothesis 3: Positive social support will moderate the relationship between child abuse and substance use, as well as the relationship between depressive symptoms and substance use.

The analytic research model is depicted in Figure 1.

Figure 1.

Conceptual research model.

Method

Study Sample

Drawn from a larger study of the social context of substance use and sexual risk, participants in this study were 445 homeless women who were randomly sampled and interviewed in temporary shelter settings in the central region of Los Angeles County as part of a larger study of the social context of substance use and sexual risk (Wenzel et al., 2009). The study aimed to reach as many representative living settings as possible following previous studies (Elliott, Golinelli, Hambarsoomian, Perlman, & Wenzel, 2006; Tucker, Wenzel, Straus, Ryan, & Golinelli, 2005). As a result, the living settings of homeless women for eligible facilities included emergency shelters, transitional living facilities, detox centers, rehabilitation centers, mental health facilities, and HIV/AIDS transitional homes. Living settings that did not serve homeless women targeted in this study were excluded. Stratified random sampling was used to select women from 51 eligible facilities, with shelters serving as sampling strata. The interviews were conducted between June 2007 and March 2008. A strict proportionate-to-size stratified random sample would have been overly burdensome at larger facilities; a modified approach was used and corrected with sampling weights.

Women were eligible to participate if they were (a) at least 18 years old, (b) had vaginal or anal sex with a male partner during the previous 6 months, (c) able to speak and understand English, and (d) did not have significant cognitive impairment. Of 472 women deemed eligible, 451 women were interviewed. Of these 451 women, six women were excluded because five women reported only oral sex with a partner, and one woman did not complete the questionnaire. Eligible women completed a personal network survey consisting of four modules; further information is available elsewhere (Green et al., 2012; Wenzel et al., 2009). Computer-assisted, face-to-face interviews were completed with 445 women, lasting approximately 1 hr and 15 min and conducted by trained female interviewers. Each woman who completed the interview was paid US$20. The final sample size was 445 women, and the response/completion rate was 94% (445 of 472). The research protocol was approved by the Institutional Review Board of the RAND Corporation.

Measures

Substance use was assessed by measuring heavy drinking, marijuana use, and crack cocaine use during the previous 6 months. Heavy drinking was assessed by asking women how often during the previous 6 months they had four or more drinks containing any kind of alcohol within a 2-hr period (0 = not at all to 9 = every day). Drug use was assessed by measuring the frequency of marijuana and crack cocaine use during the previous 6 months (0 = not at all to 9 = every day). Frequency-of-use items have been successfully employed in interviews with homeless women to assess substance use during a 6-month period (Nyamathi, Leake, & Gelberg, 2000; Tucker et al., 2005). The 10 possible response options for each substance were transformed into the number of days of heavy drinking and drug use out of 180 possible days of use (e.g., not at all was translated as zero days, and every day was translated as 180 days). Each drug had clear binomial distributions, permitting the use of binomial logistic regression to model the probability of substance use (i.e., alcohol, marijuana, or crack cocaine) on any given day during the previous 6 months.

History of childhood abuse was measured by two items asking if women experienced physical or sexual abuse by parents or other adults before the age of 18. The items used in this study were based on items from the Conflict Tactics Scale (Straus, Hamby, Boney-McCoy, & Sugarman, 1996), the National Women’s Study (Kilpatrick, Edmunds, & Seymour, 1992), and the Psychological Maltreatment of Women Survey (Tolman, 1999) and also have been used in previous studies involving homeless women (Golinelli, Longshore, & Wenzel, 2009). The physical abuse item indicated whether parents or other adults hit respondents with a fist or other objects, kicked them, beat them up, choked them, burned them, or assaulted them with a knife or gun. The sexual abuse item indicated whether parents or other adults touched the respondents in a sexual way, made them do something sexual, or made them have sex.

Depression was assessed using a two-item screening instrument tested in previous research (Rost, Burnam, & Smith, 1993) and used in a previous study with homeless people (Wenzel et al., 2012). The questions were (a) Have you had 2 years or more in your life when you felt depressed or sad most days, even if you felt okay sometimes? (b) In the past year, have you had 2 weeks or more during which you felt sad, blue, or depressed or when you lost all interest or pleasure in things that you usually cared about or enjoyed? Responses were dichotomized as in previous studies (Rost et al., 1993; Wenzel et al., 2009).

Positive social support was determined by personal network characteristics. First, women were asked to name 20 individuals aged 18 years or older who they knew, who knew them, and with whom they had contact during the previous year. Second, women were asked about support exchange and closeness among the network members they named. Finally, women were asked whether pairs of network members knew each other and how often they interacted. Detailed information can be found in Wenzel et al. (2009). The presence of a positive social support network was determined based on several questions regarding network members who were relatives to participants, such as blood relatives or relatives through marriage but excluding spouses or partners. Women were asked how often during the previous 6 months each relative network member was available to provide advice or information to help solve a problem (informational support) or to provide needed food, money, clothes, or housing (tangible support). In addition, women were asked whether they felt emotionally close to those related network members. The number of relative network members who provided any social supports (emotional closeness, tangible support, and informational support) was used as a measure of positive social support.

Sociodemographic measures included age, race/ethnicity, education level, employment status, and income. Duration of homelessness was added as a covariate. This study included two variables, religious commitment and the number of peer network members who used alcohol/drugs, based on concepts in social control and social learning theory, which have been proposed as the main competing theories to GST (Agnew, 1992; Jang & Rhodes, 2012). The influence of peers who use alcohol or drugs on an individual’s substance use has been widely studied as a social learning mechanism (Agnew, 1992; Jang & Rhodes, 2012). A measure of religiosity was included by using a single item: “Currently, how important is religion or spirituality in your daily life?”

Analytic Strategies

This study used logistic regressions to examine the relationship between child abuse and substance use, as well as the mediating role of depression and the moderating role of positive social support. Hypothesis 1 tested physical and sexual abuse as independent variables and substance use as a dependent variable. Each independent variable and dependent variable was included in separate logistic regression models.

Hypothesis 2 tested a series of logistic regressions to examine the mediating role of depression. This study followed the recommendations of Baron and Kenny (1986) regarding examination of mediators. It should be noted that this approach has been challenged by some statisticians. Zhao, Lynch, and Chen (2010) noted there is “no need for an effect to be mediated” (p. 199). Because the mediator in this study (depressive symptoms) is binary, examining its mediating effect requires a specific procedure to rescale coefficients. Following MacKinnon and Dwyer’s (1993) recommendation, this study rescaled (restandardized) coefficients before computing indirect effects produced by the mediating factor (depressive symptoms). In logistic regression, a variable has a different scale when it is used as a predictor than when it is used as an outcome variable (MacKinnon & Dwyer, 1993). To make the coefficients comparable across the equations, MacKinnon and Dwyer (1993) proposed a series of equations to compute standardized coefficients. The statistical significance of the mediating effect of depressive symptoms was tested with the Sobel test following MacKinnon and Dwyer’s (1993) recommendation.

Hypothesis 3 was examined by creating three interaction variables to test the moderating effect of positive social support. Previous research examined conditioning factors by generating interactions terms and interpreting the results as moderating effects (Piquero & Sealock, 2000). This approach was recently recommended by Agnew (2009). Based on that recommendation, two interaction variables, physical child abuse victimization × positive social support and sexual child abuse victimization × positive social support, were created. An additional interaction variable, depressive symptoms × positive social support, was generated to test the mediating effect of positive social support on the relationship between depressive symptoms and substance use among homeless women. Sampling weight was applied to all statistical procedure, and all analyses were conducted using STATA 12.0.

Results

Descriptive Characteristics

Table 1 shows that African Americans were the largest racial/ethnic group, followed by Whites and Hispanics. More than 60% of women reported earning at least a high school diploma or general educational development (GED), and about one quarter of women reported being employed. Slightly more than half of the respondents (50.4%) reported being physical abused during childhood; 43.3% reported sexual victimization during childhood. A majority of women (73.5%) screened positive for depressive symptoms. The average number of related network members who provided any positive social support (e.g., emotional closeness, tangible support, and information) during the previous 6 months was 4.82 out of 20 network members with whom participants had contact during the previous 12 months.

Table 1.

Sample Characteristics, Violent Victimization, Depression, Social Support, and Substance Use (N = 445).

Variable	%	M	SD
Individual characteristics
Age		36.56	11.10
Race/ethnicity
African American	40.17
Hispanic	22.77
White	25.86
Native American/Alaskan Native	2.17
Asian/Pacific Islander	1.38
Other/multiracial	7.65
High school graduate/GED	66.85
Employed	25.84
Monthly income		US$571.94	US$744.61
Months spent homeless		31.36	60.21
Nights at current shelter		95.40	110.01
Religiosity		3.62	0.70
Network members who use alcohol/drugs		4.09	4.07
Child abuse victimization
Physical abuse	50.40
Sexual abuse	43.30
Depressive symptoms	73.54
Relative network members who provided social support		4.82	4.07
Substance use
Heavy drinking	27.04
Marijuana	30.43
Crack cocaine	25.25

Note. GED = general educational development.

Hypothesis 1: Effect of Child Abuse Victimization on Substance Use

The association between child abuse victimization and substance use was examined through a series of logistic regressions. Contrary to hypotheses, there was no significant direct association between child abuse victimization (physical and sexual) and substance use (heavy drinking, marijuana, and crack cocaine). Several covariates were significantly associated with substance use among women. Model 1 (see Table 2) examined the association between physical child abuse and substance use; monthly income (odds ratio [OR] = 0.99, 95% confidence interval [CI = 0.99, 0.99]), nights spent at the current shelter (OR = 0.99, 95% CI = [0.98, 0.99]), and number of network members who used alcohol (OR = 1.19, 95% CI = [1.11, 1.27]) were associated with heavy drinking among respondents. Being Hispanic (OR = 0.43, 95% CI = [0.22, 0.83]), nights spent at the current shelter (OR = 0.99, 95% CI = [0.98, 0.99]), and number of network members who used alcohol and/or drugs (OR = 1.18, 95% CI = [1.10, 1.25]) were significantly associated with marijuana use. Number of nights spent at the current shelter (OR = 0.99, 95% CI = [0.97, 0.99]) and having network members who used alcohol and/or drugs (OR = 1.17, 95% CI = [1.10, 1.24]) were also significantly associated with crack cocaine use, as were age (OR = 1.06, 95% CI = [1.00, 1.05]) and employment status (OR = 0.41, 95% CI = 0.18, 0.89]). Model 2 (see Table 2) analyzed the relationship between child sexual abuse victimization and substance use. The associations between covariates and substance use were similar to Model 1. Although physical and sexual abuse victimization were not directly associated with substance use, this does not mean there is no need to test for mediating effects. Recent statisticians, including Zhao et al. (2010), have argued that the first requirement in Baron and Kenny’s (1986) procedure, that the independent variable should significantly affect the dependent variable, may not be needed to test for mediating effects. Therefore, subsequent analyses tested the mediating effect of depressive symptoms.

Table 2.

Multivariate Logistic Regression of Substance Use on Childhood Abuse.

Variable	Physical Abuse			Sexual Abuse
	Heavy Drinking	Marijuana Use	Cocaine Use	Heavy Drinking	Marijuana Use	Cocaine Use
	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]
Age	1.01 [0.99, 1.04]	0.98 [0.96, 1.01]	1.06 [1.00, 1.05]*	1.02 [1.00, 1.04]	0.99 [0.96, 1.01]	1.02 [1.00, 1.05]*
African American	0.68 [0.37, 1.26]	0.79 [0.44, 1.41]	1.01 [0.57, 1.81]	0.68 [0.37, 1.24]	0.78 [0.44, 1.40]	0.95 [0.53, 1.72]
Hispanic	1.41 [0.76, 2.61]	0.43 [0.22, 0.83]*	0.71 [0.35, 1.46]	1.38 [0.75, 2.55]	0.43 [0.22, 0.83]*	0.69 [0.34, 1.41]
High school graduate/GED	0.77 [0.46, 1.29]	0.96 [0.57, 1.60]	0.75 [0.43, 1.31]	0.78 [0.46, 1.30]	0.98 [0.58, 1.64]	0.77 [0.43, 1.37]
Employed	1.01 [0.53, 1.90]	0.84 [0.45, 1.55]	0.41 [0.18, 0.89]*	1.00 [0.53, 1.91]	0.83 [0.44, 1.53]	0.40 [0.18, 0.89]*
Monthly income	0.99 [0.99, 0.99]**	1.00 [0.99, 1.00]	1.00 [1.00, 1.00]	0.99 [0.99, 0.99]**	0.99 [1.00, 1.00]	0.99 [0.99, 1.00]
Months spent homeless	1.00 [1.00, 1.01]	1.00 [1.00, 1.01]	1.00 [1.00, 1.00]	0.99 [0.99, 1.00]	0.99 [1.00, 1.00]	1.00 [1.00, 1.00]
Nights at current shelter	0.99 [0.98, 0.99]**	0.99 [0.98, 0.99]***	0.99 [0.97, 0.99]**	0.99 [0.99, .099]**	0.99 [0.99, 0.99]***	0.99 [0.98, .99]***
Religiosity	0.86 [0.62, 1.21]	1.15 [0.82, 1.62]	0.99 [0.71, 1.40]	0.87 [0.63, 1.21]	1.16 [0.83, 1.63]	0.98 [0.70, 1.37]
Network members who use alcohol/drugs	1.19 [1.11, 1.27]***	1.18 [1.10, 1.25]***	1.17 [1.10, 1.24]***	1.19 [1.12, 1.27]***	1.18 [1.11, 1.25]***	1.17 [1.09, 1.24]***
Childhood physical abuse	1.47 [0.88, 2.47]	0.91 [0.54, 1.51]	1.08 [0.64, 1.82]
Childhood sexual abuse				1.19 [0.72, 1.99]	0.85 [0.51, 1.40]	1.38 [0.82, 2.32]

Note. OR = Odds ratio; CI = confidence interval; GED = general educational development.

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

Hypothesis 2: The Mediating Effect of Depressive Symptoms

First, the relationship between physical and sexual abuse victimization and depressive symptoms (mediator) was examined. Table 3 shows that both physical (OR = 1.98, 95% CI = [1.31, 3.00]) and sexual (OR = 2.08, 95% CI = [1.34, 3.23]) abuse victimization were significantly associated with depressive symptoms. No covariates were significantly associated with depressive symptoms among participants. Subsequently, as shown in Table 4, the independent variables (physical and sexual child abuse victimization) and the potential mediator (depressive symptoms) were included in multiple logistic regressions. The results showed that depression did not mediate the association between physical or sexual child abuse victimization and marijuana use. In contrast, depressive symptoms were significantly associated with heavy drinking and crack cocaine use in both Model 1 and Model 2. Given these results, it was expected that depressive symptoms would have a mediating effect on the link between both physical and sexual child abuse victimization and heavy drinking and crack cocaine use. To test the significance of this potential mediating effect, bootstrapping analysis was conducted for each mediating effect with rescaled coefficients for the binary mediator and outcome (MacKinnon & Dwyer, 1993). As shown in Table 5, Sobel test results showed that the effects of depressive symptoms significantly mediated the links between physical child abuse victimization and heavy drinking (z = 2.01, p < .01). Depressive symptoms also significantly mediated the link between sexual child abuse and heavy drinking (z = 2.16, p < .01) and crack cocaine use (z = 2.31, p < .01).

Table 3.

Multivariate Logistic Regression of Depressive Symptoms on Childhood Abuse.

Variable	Physical Abuse	Sexual Abuse
Variable	OR [95% CI]	OR [95% CI]
Age	1.01 [0.98, 1.02]	1.00 [0.98, 1.03]
African American	0.80 [0.48, 1.31]	0.76 [0.46, 1.26]
Hispanic	0.76 [0.44, 1.33]	0.75 [0.43, 1.32]
High school graduate/GED	0.84 [0.52, 1.32]	0.87 [0.54, 1.29]
Employed	0.94 [0.56, 1.61]	0.97 [0.58, 1.63]
Monthly income	1.00 [0.99, 1.00]	1.00 [1.00, 1.00]
Months spent homeless	1.00 [0.99, 1.00]	1.00 [1.00, 1.00]
Nights at current shelter	1.00 [1.00, 1.00]	1.00 [1.00, 1.00]
Religiosity	0.76 [0.55, 1.04]	0.74 [0.54, 1.01]
Childhood physical abuse	1.98 [1.31, 3.00]**
Childhood sexual abuse		2.08 [1.34, 3.23]**

Note. OR = Odds ratio; CI = confidence interval; GED = general educational development.

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

Table 4.

Multivariate Logistic Regression of the Mediating Role of Depressive Symptoms on Substance Use and Childhood Abuse.

Variable	Physical Abuse			Sexual Abuse
Variable	Heavy DrinkingOR [95% CI]	Marijuana UseOR [95% CI]	Cocaine UseOR [95% CI]	Heavy DrinkingOR [95% CI]	Marijuana UseOR [95% CI]	Cocaine UseOR [95% CI]
Age	1.01 [0.99, 1.04]	0.98 [0.96, 1.01]	1.03 [1.00, 1.05]*	1.01 [0.99, 1.04]	0.98 [0.96, 1.01]	1.03 [1.00, 1.05]*
African American	0.69 [0.37, 1.28]	0.80 [0.45, 1.42]	1.04 [0.58, 1.86]	0.70 [0.37, 1.29]	0.79 [0.43, 1.42]	0.98 [0.54, 1.78]
Hispanic	1.43 [0.76, 2.70]	0.43 [0.22, 0.83]*	0.71 [0.34, 1.49]	1.42 [1.75, 2.69]	0.43 [0.22, 0.83]*	0.70 [0.34, 1.45]
High school graduate/GED	0.79 [0.47, 1.35]	0.97 [0.58, 1.63]	0.77 [0.44, 1.37]	0.80 [0.47, 1.36]	0.99 [0.59, 1.68]	0.80 [0.45, 1.43]
Employed	1.01 [0.53, 1.92]	0.84 [0.45, 1.55]	0.41 [0.18, 0.89]	0.99 [0.52, 1.90]	0.83 [0.45, 1.54]	0.41 [0.19, 0.88]*
Monthly income	0.99 [0.99, 0.99]**	0.99 [0.99, 1.00]	0.99 [0.99, 1.00]	0.99 [0.99, 0.99]**	0.99 [0.99, 1.00]	0.99 [0.99, 1.00]
Months spent homeless	1.00 [0.99, 1.00]	1.00 [0.99, 1.00]	1.00 [0.99, 1.00]	0.99 [0.98, 1.00]	1.00 [0.99, 1.00]	1.00 [0.99, 1.00]
Nights at current shelter	0.99 [0.99, 1.00]**	0.99 [0.98, 0.99]***	0.99 [0.98, 0.99]**	0.99 [0.99, 0.99]**	0.99 [0.98, 0.99]***	0.99 [0.98, 0.99]**
Religiosity	0.90 [0.65, 1.25]	1.17 [0.83, 1.64]	1.04 [0.74, 1.47]	0.91 [0.65, 1.27]	1.17 [0.84, 1.64]	1.02 [0.73, 1.44]
Network members who use alcohol/drugs	1.18 [1.11, 1.26]***	1.18 [1.11, 1.25]***	1.16 [1.09, 1.24]***	1.19 [1.11, 1.26]***	1.18 [1.11, 1.25]***	1.16 [1.09, 1.23]***
Childhood physical abuse	1.33 [0.78, 2.27]	0.88 [0.53, 1.49]	0.96 [0.56, 1.64]
Childhood sexual abuse				1.05 [0.62, 1.82]	0.82 [0.48, 1.38]	1.24 [0.73, 2.13]
Depressive symptoms	2.15 [1.24, 3.70]**	1.17 [0.67, 2.03]	2.14 [1.18, 3.87]**	2.28 [1.31, 3.98]**	1.19 [0.68, 2.11]	1.95 [1.08, 3.53]**

Note. OR = Odds ratio; CI = confidence interval; GED = general educational development.

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

Table 5.

Test of Significance of Indirect Effects of Depressive Symptoms Using Bootstrapping.

Abuse Type	Path	z
Physical	PA → DS → heavy drinking	2.01*
	PA → DS → marijuana	0.45
	PA → DS → crack cocaine	1.90
Sexual	SA → DS → heavy drinking	2.16*
	SA → DS → marijuana	0.61
	SA → DS → crack cocaine	2.31*

Note. PA = Physical abuse; DS = depressive symptoms; SA = sexual abuse.

p < .01.

Hypothesis 3: The Moderating Effect of Positive Social Support

To analyze Hypothesis 3, three interaction variables to test the moderating effect of positive social support on child abuse and substance use among the women were created. The result is depicted in Table 6. The results show that the moderating effects of positive social support worked differently by type of child abuse and substance use. Positive social support significantly moderated the effect of physical (OR = 0.85, 95% CI = [0.79, 0.93]) and sexual (OR = 0.80, 95% CI = [0.67, 0.94]) child abuse only on heavy drinking. This indicates that the effect of physical and sexual child abuse victimization on heavy drinking decreased as positive social support increased. Significant moderating effects of positive social support were not detected for marijuana and crack cocaine use in this study.

Table 6.

Moderating Effects of Social Support From Relatives on Substance Use.

Variable	Heavy Drinking	Marijuana Use	Cocaine Use
Variable	OR [95% CI]	OR [95% CI]	OR [95% CI]
Age	1.02 [0.99, 1.05]	0.98 [0.95, 1.00]	1.02 [0.99, 1.05]
African American	0.63 [0.30, 1.30]	0.73 [0.42, 1.37]	1.04 [0.55, 1.97]
Hispanic	1.44 [0.76, 2.74]	0.42 [0.20, 0.82]*	0.72 [0.35, 1.49]
High school graduate/GED	0.83 [0.48, 1.46]	0.99 [0.56, 1.65]	0.76 [0.44, 1.32]
Employed	0.82 [0.41, 1.64]	0.77 [0.41, 1.50]	0.39 [0.18, 0.92]*
Monthly income	0.99 [0.99, 0.99]**	1.00 [0.99, 1.00]	0.99 [0.99, 1.00]
Months spent homeless	0.99 [0.98, 1.01]	1.00 [0.99, 1.00]	1.00 [0.99, 1.01]
Nights at current shelter	0.99 [0.99, 0.99]**	0.99 [0.98, 0.99]***	0.99 [0.98, 0.99]**
Religiosity	0.91 [0.64, 1.28]	1.19 [0.83, 1.70]	1.02 [0.80, 1.47]
Network members who use alcohol/drugs	1.22 [1.14, 1.30]***	1.18 [1.11, 1.25]***	1.16 [1.09, 1.24]***
Childhood physical abuse (A)	1.32 [0.72, 4.28]	1.71 [0.29, 1.69]	0.69 [0.28, 1.73]
Childhood sexual abuse (B)	2.62 [0.88, 6.11]	1.99 [0.40, 2.46]	1.89 [0.75, 4.74]
Depressive symptoms (C)	1.83 [0.77, 4.36]	0.67 [0.38, 2.12]	1.86 [0.69, 4.94]
Network members who provided social support (D)	0.92 [0.80, 1.05]	0.91 [0.80, 1.03]	0.98 [0.80, 1.10]
A × D	0.85 [0.79, 0.93]**	1.06 [0.92, 1.23]	1.02 [0.81, 1.26]
B × D	0.80 [0.67, 0.94]*	1.01 [0.86, 1.16]	0.94 [0.80, 1.15]
C × D	1.06 [0.91, 1.22]	1.14 [0.94, 1.24]	1.03 [0.86, 1.20]

Note. OR = Odds ratio; CI = confidence interval; GED = general educational development.

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

Discussion

The purpose of this study was to determine whether the GST (Agnew, 1992) theoretical framework explained substance use among homeless women. This study focused on physical and sexual child abuse victimization as factors that may be related to substance use (heavy drinking, marijuana, and crack cocaine) among homeless women. Although previous studies consistently indicated that childhood victimization may have long-term correlates with substance use in adulthood among the general population (McClellan, Farabee, & Crouch, 1997; Meade, Kershaw, Hansen, & Sikkema, 2009; Shin, McDonald, & Conley, 2018), the results on the association were inconsistent among homeless women. This suggests the need to consider other factors on the path between child abuse victimization and adulthood substance use. GST, which focuses on the dynamic of factors associated with strains and deviant behaviors including substance use, can be a useful tool to examine the factors around child abuse victimization and adulthood substance use among these women.

Overall research results indicated that homeless women who experienced child abuse in the past are more likely to show depressive symptoms and high prevalence of substance use. Practitioners working with homeless women with substance use should assess their child abuse experience and depressive symptoms and understand the dynamic between those factors to plan effective substance use intervention strategies. More specifically, three major assumptions of GST were tested with a randomly selected sample of homeless women in temporary shelter. The three hypotheses of this study involved the association between child abuse victimization and substance use, the mediating effect of depressive symptoms on the relationship between child abuse victimization and substance use, and the moderating effect of positive social support on the relationship between child abuse victimization and substance use and the relationship between depressive symptoms and substance use.

Consistent with previous research (Upshur et al., 2017), descriptive results showed that the homeless women in this study showed a high prevalence rate of heavy drinking, marijuana use, and crack cocaine use. The empirical tests of GST assumptions showed that the theory was only partially supported. Hypothesis 1 was not supported. Multivariate logistic regressions indicated that neither physical abuse nor sexual abuse were significantly associated with any form of substance use. Hypothesis 2 was partially supported. Child abuse experience of women was indirectly associated with current substance use. Depressive symptoms of women fully mediated the four associations (physical abuse and heavy drinking, physical abuse and crack cocaine use, sexual child abuse victimization and heavy drinking, and sexual child abuse victimization and crack cocaine use). Women with physical and sexual abuse experience in childhood were more likely to experience depressive symptoms, and depressive symptoms may amplify the likelihood of heavy drinking and crack cocaine. Although there was no significant direct effect of child abuse victimization on substance use, the results indicated the effects of child abuse victimization may be mediated by depression, especially in terms of heavy drinking and crack cocaine use among homeless women. Practitioners may focus on addressing depressive symptoms as a strategy to reduce substance use (especially heavy drinking and crack cocaine use) among homeless women who have a history of physical and sexual abuse during childhood. Also, results showed that child abuse victimization was associated with depressive symptoms during adulthood, which may be associated with substance use. This may suggest that examining a client’s history of abuse during childhood may help practitioners gain a better understanding of the causes of mental illness, depressive symptoms, and substance use. Hypothesis 3 was partially supported. Positive social support was shown to have potential usefulness in terms of reducing the effect of physical and sexual child abuse victimization on heavy drinking. This suggests that the effect of physical and sexual abuse on heavy drinking may decrease as positive social support increases. Increasing positive social support, especially from relatives, may attenuate the likelihood of heavy drinking among homeless women who have a history of child abuse victimization.

Overall, the important contribution of the current study can be described as follows: First, this study addressed the previous inconsistent results of the association between child abuse victimization and substance use among homeless women by adopting the theoretical framework of GST, which has been less applied to homeless women despite its usefulness in addressing many types of behavioral problems among homeless women. As Jones et al. (2018) described, GST can be a useful tool for women whose life conditions can be conditioned by multiple layers of disadvantages and adverse life events. Second, by showing the important role of depressive symptoms and social supports of homeless women, the results of this study indicate that the path between child abuse victimization and substance use is complex. Based on the results, some practical implications can be discussed. Practitioners working with homeless women should be informed that child abuse victimization that may not be observed at the moment can have long-term effects on the lives of homeless women even after they become adults. Trauma-informed care is necessary to address negative experiences in childhood. Considering the life condition of homeless people, practitioners provide trauma-informed care (Hopper, Bassuk, & Olivet, 2010). The results suggest that improving women’s positive social supports may play a significant role in reducing substance use, particularly heavy drinking, among those who were physically or sexually abused in childhood. However, it is premature to conclude positive social supports are effective only on heavy drinking. Further research should identify the moderators that are effective on substance abuse and sexual abuse.

There were several limitations to this study that are worth noting. Strains (child abuse victimization) were objective. As Agnew (2001) noted, subjective measures of how individuals feel about objective strains should be included to accurately assess the role of strains on deviant behaviors. This study focused on depressive symptoms as representative of negative emotion. Other negative emotions such as anger, anxiety, and posttrauma symptoms may lead to substance use. Future studies should include other negative emotions as possible mediators. This study did not reflect the quality of positive social support received from relative network members. In future studies, rigorous measurements to examine negative emotions need to be used to assess more accurately the status of emotionality.

In sum, this study explored the association between child abuse and substance use using GST, which provided a practical framework to understand substance use among homeless women. With more accurate measurements and advanced methods, researchers may identify additional mediators and moderators and clarify the link between child abuse victimization and substance use. Evidence from those efforts may inform the development of highly effective interventions to reduce substance use problems among homeless women.

Footnotes

Acknowledgements

We thank Dr. Daniela Golinelli for her contributions on sample design, and Dr. David Kennedy and Dr. Hank Green for their contributions on the social network interview.

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: This research was supported by the grant number, R01HD059307.

ORCID iD

Ahyoung Song

Author Biographies

Ahyoung Song, PhD, has participated in multiple projects addressing interpersonal violence and risky health behaviors among women including homeless women. Her current research interests are homelessness, housing policies for vulnerable population and community-based participatory research in social work. She received her PhD at the University of Southern California and is now affiliated with the Gachon University, South Korea.

Suzanne L. Wenzel, PhD, has devoted much of her career to understand and address health-related needs of vulnerable populations, particularly individuals experiencing homelessness in urban communities. She serves as the director of the School of Social Work research cluster on Homelessness, Housing, and Social Environment.

Yusun Cho received her PhD at the University of Southern California in Public Policy and Management. Her research interests include gender and minority policy issues, nonprofit and public management, social service delivery, collaborative governance, and social network analysis.

References

Agnew

(1992). Foundation for a general strain theory of crime and delinquency. Criminology, 30, 47-88. doi:10.1111/j.1745-9125.1992.tb01093.x

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 & Delinquency, 38, 319-361. doi:10.1177/0022427801038004001

Agnew

(2006). General strain theory: Current status and directions for further research. In Cullen

F. T.

Wright

J. P.

Blevins

K. R.

(Eds.), Taking stock: The status of criminological theory: Advances in criminological theory (Vol. 15, pp. 101-126). New Brunswick, NJ: Transaction.

Agnew

(2009). General strain theory. In Krohn

M. D.

Lizotte

A. J.

Hall

G. P.

(Eds.), Handbook on crime and deviance (pp. 169-186). New York, NY: Springer.

Agnew

White

H. R.

(1992). An empirical test of general strain theory. Criminology, 30, 475-500. doi:10.1111/j.1745-9125.1992.tb01113.x

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. doi:10.1037/0022-3514.51.6.1173

Bassuk

E. L.

Weinreb

L. F.

Buckner

J. C.

Browne

Salomon

Bassuk

S. S.

(1996). The characteristics and needs of sheltered homeless and low-income housed mothers. Journal of the American Medical Association, 276, 640-646. doi:10.1001/jama.1996.03540080062031

Broidy

L. M.

Agnew

(1997). Gender and crime: A general strain theory perspective. Journal of Research in Crime & Delinquency, 34, 275-306. doi:10.1177/0022427897034003001

Browne

(1993). Family violence and homelessness: The relevance of trauma histories in the lives of homeless women. American Journal of Orthopsychiatry, 63, 370-384. doi:10.1037/h0079444

10.

Chen

Tyler

K. A.

Whitbeck

L. B.

Hoyt

D. R.

(2004). Early sexual abuse, street adversity, and drug use among female homeless and runaway adolescents in the Midwest. Journal of Drug Issues, 34, 1-21. doi:10.1177/002204260403400101

11.

Elliott

M. N.

Golinelli

Hambarsoomian

Perlman

Wenzel

S. L.

(2006). Sampling with field burden constraints: An application to sheltered homeless and low-income housed women. Field Methods, 18, 43-58. doi:10.1177/1525822X05284014

12.

Folger

S. F.

Wright

M. O.

(2013). Altering risk following child maltreatment: Family and friend support as protective factors. Journal of Family Violence, 28, 325-337. doi:10.1007/s10896-013-9510-4

13.

Galaif

E. R.

Nyamathi

A. M.

Stein

J. A.

(1999). Psychosocial predictors of current drug use, drug problems, and physical drug dependence in homeless women. Addictive Behaviors, 24, 801-814. doi:10.1016/S0306-4603(99)00038-6

14.

Golinelli

Longshore

Wenzel

S. L.

(2009). Substance use and intimate partner violence: Clarifying the relevance of women’s use and partners’ use. Journal of Behavioral Health Services & Research, 36, 199-211. doi:10.1007/s11414-008-9114-6

15.

Green

H. D.

Jr. Tucker

J. S.

Wenzel

S. L.

Golinelli

Kennedy

D. P.

Ryan

G. W.

Zhou

A. J.

(2012). Association of childhood abuse with homeless women’s social networks. Child Abuse & Neglect, 36, 21-31. doi:10.1016/j.chiabu.2011.07.005

16.

Grothoff

G. E.

Kempf-Leonard

Mullins

(2014). Gender and juvenile drug abuse: A general strain theory perspective. Women & Criminal Justice, 24, 22-43. doi:10.1080/08974454.2013.842519

17.

Herrenkohl

T. I.

Hong

Klika

J. B.

Herrenkohl

R. C.

Russo

M. J.

(2013). Developmental impacts of child abuse and neglect related to adult mental health, substance use, and physical health. Journal of Family Violence, 28, 191-199. doi:10.1007/s10896-012-9474-9

18.

Hopper

E. K.

Bassuk

E. L.

Olivet

(2010). Shelter from the storm: Trauma-informed care in homelessness service settings. The Open Services and Policy Journal, 3, 80-100. doi:10.2174/1874924001003010080

19.

Jang

S. J.

(2007). Gender differences in strain, negative emotions, and coping behaviors: A general strain theory approach. Justice Quarterly, 24, 523-553. doi:10.1080/07418820701485486

20.

Jang

S. J.

Rhodes

J. R.

(2012). General strain and non-strain theories: A study of crime in emerging adulthood. Journal of Criminal Justice, 40, 176-186. doi:10.1016/j.jcrimjus.2011.09.003

21.

Jasinski

J. L.

Wesely

J. K.

Mustaine

Wright

J. D.

(2005). The experience of violence in the lives of homeless women: A research report. Washington, DC: National Institute of Justice.

22.

Jones

M. S.

Worthen

M. G.

Sharp

S. F.

McLeod

D. A.

(2018). Bruised inside out: The adverse and abusive life histories of incarcerated women as pathways to PTSD and illicit drug use. Justice Quarterly, 35(6), 1004-1029. doi:10.1080/07418825.2017.1355009

23.

Kaufman

J. M.

(2009). Gendered responses to serious strain: The argument for a general strain theory of deviance. Justice Quarterly, 26, 410-444. doi:10.1080/07418820802427866

24.

Kilpatrick

D. G.

Edmunds

C. N.

Seymour

A. K.

(1992). Rape in America: A report to the nation. Arlington, VA: National Victim Center.

25.

Kort-Butler

L. A.

(2010). Experienced and vicarious victimization: Do social support and self-esteem prevent delinquent responses? Journal of Criminal Justice, 38, 496-505. doi:10.1016/j.jcrimjus.2010.04.019

26.

MacKinnon

D. P.

Dwyer

J. H.

(1993). Estimating mediated effects in prevention studies. Evaluation Review, 17, 144-158. doi:10.1177/0193841X9301700202

27.

Manasse

M. E.

Ganem

N. M.

(2009). Victimization as a cause of delinquency: The role of depression and gender. Journal of Criminal Justice, 37, 371-378. doi:10.1016/j.jcrimjus.2009.06.004

28.

McClellan

D. S.

Farabee

Crouch

B. M.

(1997). Early victimization, drug use, and criminality. Criminal Justice and Behavior, 24, 455-476. doi:10.1177/0093854897024004004

29.

Meade

C. S.

Kershaw

T. S.

Hansen

N. B.

Sikkema

K. J.

(2009). Long-term correlates of childhood abuse among adults with severe mental illness: Adult victimization, substance abuse, and HIV sexual risk behavior. AIDS and Behavior, 13, 207-216. doi:10.1007/s10461-007-9326-4

30.

Moon

Morash

(2017). Gender and general strain theory: A comparison of strains, mediating, and moderating effects explaining three types of delinquency. Youth & Society, 49, 484-504. doi:10.1177/0044118X14541877

31.

Najdowski

C. J.

Ullman

S. E.

(2011). The effects of revictimization on coping and depression in female sexual assault victims. Journal of Traumatic Stress, 24, 218-221. doi:10.1002/jts.20610

32.

Nyamathi

A. M.

Leake

Gelberg

(2000). Sheltered versus nonsheltered homeless women: Differences in heath, behavior, victimization, and utilization of care. Journal of General Internal Medicine, 15, 565-572. doi:10.1046/j.1525-1497.2000.07007.x

33.

Piquero

N. L.

Sealock

M. D.

(2000). Generalizing general strain theory: An examination of an offending population. Justice Quarterly, 17, 449-484. doi:10.1080/07418820000094631

34.

Piquero

N. L.

Sealock

M. D.

(2004). Gender and general strain theory: A preliminary test of Broidy and Agnew’s gender/GST hypotheses. Justice Quarterly, 21, 125-158. doi:10.1080/07418820400095761

35.

Ross

L. E.

Dennis

C.-L.

(2009). The prevalence of postpartum depression among women with substance use, an abuse history, or chronic illness: A systematic review. Journal of Women’s Health, 18, 475-486. doi:10.1089/jwh.2008.0953

36.

Rost

Burnam

M. A.

Smith

G. R.

(1993). Development of screeners for depressive disorders and substance disorder history. Medical Care, 31, 189-200. doi:10.1097/00005650-199303000-00001

37.

Shin

S. H.

McDonald

S. E.

Conley

(2018). Patterns of adverse childhood experiences and substance use among young adults: A latent class analysis. Addictive Behaviors, 78, 187-192. doi:10.1016/j.addbeh.2017.11.020

38.

Shorey

R. C.

Sherman

A. E.

Kivisto

A. J.

Elkins

S. R.

Rhatigan

D. L.

Moore

T. M.

(2011). Gender differences in depression and anxiety among victims of intimate partner violence: The moderating effect of shame proneness. Journal of Interpersonal Violence, 26, 1834-1850. doi:10.1177/0886260510372949

39.

Stein

J. A.

Dixon

E. L.

Nyamathi

A. M.

(2008). Effects of psychosocial and situational variables on substance abuse among homeless adults. Psychology of Addictive Behaviors, 22, 410-416. doi:10.1037/0893-164X.22.3.410

40.

Stein

J. A.

Leslie

M. B.

Nyamathi

(2002). Relative contributions of parent substance use and childhood maltreatment to chronic homelessness, depression, and substance abuse problems among homeless women: Mediating roles of self-esteem and abuse in adulthood. Child Abuse & Neglect, 26, 1011-1027. doi:10.1016/S0145-2134(02)00382-4

41.

Straus

M. A.

Hamby

S. L.

Boney-McCoy

Sugarman

D. B.

(1996). The revised Conflict Tactics Scales (CTS2): Development and preliminary psychometric data. Journal of Family Issues, 17, 283-316. doi:10.1177/019251396017003001

42.

Tolman

R. M.

(1999). The validation of the psychological maltreatment of women inventory. Violence and Victims, 14, 25-37.

43.

Tucker

J. S.

Wenzel

S. L.

Straus

J. B.

Ryan

G. W.

Golinelli

(2005). Experiencing interpersonal violence: Perspectives of sexually active, substance-using women living in shelters and low-income housing. Violence Against Women, 11, 1319-1340. doi:10.1177/1077801205280190

44.

Tyler

K. A.

Melander

L. A.

(2015). Child abuse, street victimization, and substance use among homeless young adults. Youth & Society, 47, 502-519. doi:10.1177/0044118X12471354

45.

Upshur

C. C.

Jenkins

Weinreb

Gelberg

Orvek

E. A.

(2017). Prevalence and predictors of substance use disorders among homeless women seeking primary care: An 11 site survey. The American Journal on Addictions, 26, 680-688. doi:10.1111/ajad.12582

46.

Wenzel

S. L.

Green

H. D.

Jr. Tucker

J. S.

Golinelli

Kennedy

D. P.

Ryan

Zhou

(2009). The social context of homeless women’s alcohol and drug use. Drug and Alcohol Dependence, 105, 16-23. doi:10.1016/j.drugalcdep.2009.05.026

47.

Wenzel

S. L.

Rhoades

Hsu

H.-T.

Golinelli

Tucker

J. S.

Kennedy

D. P.

Ewing

(2012). Behavioral health and social normative influence: Correlates of concurrent sexual partnering among heterosexually-active homeless men. AIDS and Behavior, 16, 2042-2050. doi:10.1007/s10461-011-0066-0

48.

Zhao

Lynch

J. G.

Chen

(2010). Reconsidering Baron and Kenny: Myths and truths about mediation analysis. Journal of Consumer Research, 37, 197-206. doi:10.1086/651257