Gender, coping strategies, homelessness stressors, and income generation among homeless young adults in three cities

Social Science & Medicine 135 (2015) 47e55

Contents lists available at ScienceDirect

Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed

Gender, coping strategies, homelessness stressors, and income generation among homeless young adults in three cities Kristin M. Ferguson a, *, Kimberly Bender b, Sanna J. Thompson c a

Silberman School of Social Work at Hunter College, 2180 Third Avenue, New York, NY 10035, USA University of Denver, School of Social Work, USA c University of Texas at Austin, School of Social Work, USA b

a r t i c l e i n f o

a b s t r a c t

Article history: Available online 26 April 2015

This study examined gender differences among homeless young adults' coping strategies and homelessness stressors as they relate to legal (e.g., full-time employment, selling personal possessions, selling blood/plasma) and illegal economic activity (e.g., selling drugs, theft, prostitution). A sample of 601 homeless young adults was recruited from 3 cities (Los Angeles, CA [n ¼ 200], Austin, TX [n ¼ 200], and Denver, CO [n ¼ 201]) to participate in semi-structured interviews from March 2010 to July 2011. Risk and resilience correlates of legal and illegal economic activity were analyzed using six Ordinary Least Squares regression models with the full sample and with the female and male sub-samples. In the full sample, three variables (i.e., avoidant coping, problem-focused coping, and mania) were associated with legal income generation whereas eight variables (i.e., social coping, age, arrest history, transience, peer substance use, antisocial personality disorder [ASPD], substance use disorder [SUD], and major depressive episode [MDE]) were associated with illegal economic activity. In the female sub-sample, three variables (i.e., problem-focused coping, race/ethnicity, and transience) were correlated with legal income generation whereas six variables (i.e., problem-focused coping, social coping, age, arrest history, peer substance use, and ASPD) were correlated with illegal economic activity. Among males, the model depicting legal income generation was not significant yet seven variables (i.e., social coping, age, transience, peer substance use, ASPD, SUD, and MDE) were associated with illegal economic activity. Understanding gender differences in coping strategies and economic activity might help customize interventions aimed at safe and legal income generation for this population. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Coping Employment Gender Homeless young adults Income generation Mental health Survival behavior

Research on homeless young people's employment has grown over the past decade resulting in greater knowledge of their economic activities (Ferguson et al., 2012; Gaetz and O'Grady, 2002). These young people commonly earn income via combinations of formal and informal sources, the latter of which can be legal (e.g., selling self-made items, possessions, and blood/plasma) or illegal (e.g., selling drugs, theft, and prostitution; Gaetz and O'Grady, 2002). In light of cited gender bias and segregation within the formal and informal labor markets (O'Grady and Gaetz, 2004), one limitation of extant research is the lack of attention to differential effects in how male and female homeless youth generate income. In the few studies examining the gendered nature of homeless

* Corresponding author. E-mail addresses: [email protected] (K.M. Ferguson), kimberly.bender@du. edu (K. Bender), [email protected] (S.J. Thompson). http://dx.doi.org/10.1016/j.socscimed.2015.04.028 0277-9536/© 2015 Elsevier Ltd. All rights reserved.

youth's involvement in legal and illegal economic activities, finding reveal that although male and female youth did not differ in employment status, females reported holding fewer concurrent positions and having lower incomes than males. They were also involved in different types of economic activities than males. For instance, males were more likely to report a greater number of positions at one time, higher weekly earnings, and being employed full-timedfindings that are consistent with housed youth as well (Gabriel and Schmitz, 2006). Males were also more likely to earn income via criminal activity as well as through manual-labor positions. Conversely, females were more likely to work in customer service positions and in the sex trade economy (O'Grady and Gaetz, 2004; Robinson and Baron, 2007). Although these findings highlight gender differences in homeless youth's economic activities, it remains unclear why some male and female youth engage in legal income generation whereas others choose to participate in illegal work. Prior research suggests

48

K.M. Ferguson et al. / Social Science & Medicine 135 (2015) 47e55

that homelessness stressors (e.g., food and housing insecurity, mental illness, and limited employment skills) constitute barriers to formal employment (Dachner and Tarasuk, 2002) and are associated with illicit forms of income generation (Ferguson et al., 2011). Yet many youth experiencing these stressors participate in the formal labor market (Ferguson et al., 2012; Gaetz and O'Grady, 2002; O'Grady and Gaetz, 2004) and do not engage in illegal work. What then buffers homeless young people from the stressors of homelessness to enable them to obtain and maintain safe and legal economic activity? Likewise, are males and females protected by different factors? Researchers studying the impact of homelessness strains on criminal behavior suggest that individual coping mechanisms might help explain why particular youth engage in illegal activity (Baron, 2004). It might be that the strategies male and female homeless young people use to cope with these stressors help explain differences in the types of economic activity they seek. To explore this supposition, this study used the risk and resilience framework to examine correlates of legal and illegal income generation by homeless young adults in three U.S. cities and how these factors vary by gender.

Further, mental illness and substance use are barriers to formal employment (Bond and Drake, 2008) and correlates of illegal activity among homeless youth (Baron, 2004). As individuals with mental illness (e.g., depression, mania, and personality disorders) become disconnected from mainstream services and treatment, they can be prone to illegal acts (Silver, 2000). In contrast, income generation from formal and legal informal sources can be protective. Homeless youth's involvement in formal employment is associated with various positive outcomes, such as stable housing and mental health treatment (Ferguson et al., 2012). Employment is particularly important to this population since it contributes to their identity formation, links them to conventional institutions, and provides income that facilitates economic selfsufficiency (Gaetz and O'Grady, 2002). Similarly, safe and legal informal work can provide them with daily income to meet their subsistence needs (O'Grady and Gaetz, 2004) as well as access to supportive peers and resources.

1. Theoretical framework

How homeless young people cope with adversity in ways that are adaptive or maladaptive can shape how they experience homelessness, including how they make money to survive. Literature examining coping and resiliency in this population suggests that they rely on varied coping strategies such as problem-focused coping (i.e., attempts to address the problem or stressor itself), avoidant coping (i.e., problem and emotional avoidance), and social coping (i.e., use of social support and social withdrawal; Kidd and Carroll, 2007). Prior research indicates that homeless youth who used problem-focused coping strategies experienced positive health and mental health outcomes (Unger et al., 1998), whereas use of avoidant or disengagement coping strategies was associated with greater rates of both mental illness (e.g., depression) and behavior problems (Votta and Manion, 2003) as well as suicidal ideation (Kidd and Carroll, 2007). Gender differences have also been noted in homeless youth's coping methods, with females and males using different strategies to deal with suicide risk (Kidd and Carroll, 2007). For instance, social withdrawal coping was associated with feeling trapped or helpless for females but not for males. Likewise, avoidant coping techniques (e.g., sleep) were associated with feeling trapped or helpless for females but not for males (Kidd and Carroll, 2007).

The risk and resilience framework is useful for explaining how intrapersonal and environmental risk and protective factors inhibit and promote positive youth development including the pursuit of employment. Intrapersonal and environmental risk factors increase the likelihood of problem behaviors and negative outcomes. In contrast, protective factors refer to the individual and environmental conditions that decrease the likelihood of problem behaviors or that buffer the effects of risk (Fraser et al., 1999). Risk and protective factors are important concepts for homeless young people since many of their background factors contributing to homelessness and their ensuing experiences once homeless are associated with negative outcomes. Conversely, that many of these young people left home to avoid dysfunctional environments suggests that they possess resiliency despite surrounding adversity (Rew et al., 2001). Once homeless, they often adopt survival behaviors to cope with and adapt to their daily stressors as well as to generate income to meet their subsistence needs (Greene et al., 1999). Understanding how risk and protective factors interact in male and female homeless young adults can help them better navigate homelessness and make decisions that enable them to achieve employment and make successful transitions to adulthood.

1.2. Coping as a protective factor

1.1. Risk and protective factors associated with income generation 2. Research questions One consequence of homelessness stressors and economic marginalization is the attraction to informal (often illicit) economic activities (Baron and Hartnagel, 1997; Greene et al., 1999). Often due to labor exclusion, homeless young people engage in various informal means of legal and illegal income generation (i.e., survival behaviors) in particular the longer they remain homeless (Gaetz and O'Grady, 2002). Income generated this way might be more readily available for this transient population since survival behaviors can be used in any city, require fewer commitments and relationships with institutions, and offer immediate income (Ferguson et al., 2011). Similarly, homeless youth's involvement with substance-using peers is also a correlate of informal and illegal income generation. These networks of like-minded peers endorse substance use as a coping mechanism and commonly engage in illicit behaviors. Such illegal activities often take place in unsupervised and unsafe locations, expose homeless youth to dangerous adults and peers, and increase their risk of exploitation, trauma, and victimization (Ferguson et al., 2011; O'Grady and Gaetz, 2004).

It is evident that homeless young people encounter myriad stressors that vary by gender. Their ability to cope with these stressors contributes to positive overall health, mental health, and behavioral outcomes. Despite the recognition that coping strategies are an important protective factor for this population, prior research has not examined how these strategies influence their economic activity or whether strategies differ by gender. Thus, the purpose of this study was to examine homeless young adults' risk factors (i.e., homelessness stressors) and protective factors (i.e., coping strategies) related to legal and illegal income generation as well as to determine how these correlates vary by gender. Three research questions guided this study: 1) how do males and females differ in their sources of legal and illegal income generation, 2) what coping strategies are associated with income generated from legal and illegal sources controlling for homelessness risk factors, and 3) how do these factors vary by gender?

K.M. Ferguson et al. / Social Science & Medicine 135 (2015) 47e55

3. Methods 3.1. Design and research settings This cross-sectional, comparative study of homeless young adults was conducted at three homeless youth agencies (one per city) in Los Angeles, CA; Austin, TX; and Denver, CO. Agencies were selected based on their existing relationships with researchers and their commitment to host the study. Participating agencies were multi-service, non-profit organizations that offer homeless, runaway, and at-risk young people a comprehensive system of care including street outreach, meals, shelter, health care, counseling, and educational and employment services. Human subjects' approval was received by each investigator's university. 3.2. Participants and recruitment procedures Using purposive sampling, researchers and trained research assistants in each city recruited 200 homeless young adults (ages 18e24) using similar methods (201 participants were recruited in Denver). Researchers recruited participants from among three different service programs at the three host agencies: (1) street outreach/drop-in centers (non-residential); (2) residential shortterm and mid-length shelters (30 days to up to 6 years); and (3) transitional housing (long-term housing). Recruitment took place from March 2010 to July 2011 during agencies' service hours. To participate, young adults had to (1) be 18e24 years of age; (2) have spent at least 2 weeks away from home in the month before the interview (Whitbeck, 2009); and (3) provide written informed consent. A screening form was used to determine participants' ages and length of time away from home. Agency staff made introductions between participants and interviewers, who then explained the study procedures, secured written consent, and administered the interview. 3.3. Data collection and measures Researchers across the three cities administered a 45-min, quantitative, semi-structured interview to 601 young adults within the host agencies in a private room. The interview sought information on income generation, homelessness history, criminal history, mental health, and an array of risk and protective factors. Interviewers read questions and response options aloud to participants who responded verbally. Participants were compensated with a $10.00 gift card. 3.3.1. Dependent variables: income generation via legal and illegal sources This study used two dependent variables. First, income generated via legal sources (Legal Sources) was determined by asking the young people whether they got any money or resources during the past 6 months to meet their basic needs from any one of seven legal sources of income: 1) full-time employment, 2) part-time employment, 3) temporary paid employment, 4) selling selfmade items, 5) selling bottles/cans, 6) selling clothing or other personal possessions, and 7) selling their blood/plasma (0 ¼ no, 1 ¼ yes). Responses were then summed to form a continuous variable representing the variety of legal sources from which the young people earned income during the previous 6 months (Range 0e7). The second dependent variable (Illegal Sources) was created from the interview question asking the young adults whether they got any money or resources during the past 6 months to meet their basic needs from any one of five illegal (or legally regulated) sources of income: 1) panhandling, 2) dealing drugs, 3) survival sex

49

(i.e., exchanging sexual acts for money, food, or lodging, or to meet other needs), 4) stealing, and 5) gambling (0 ¼ no, 1 ¼ yes). Responses to these five items were then added to form a continuous variable that represented the variety of illegal sources from which the young adults earned income during the prior 6 months (Range 0e5). Due to the large skewness of both original variables (Legal Sources ¼ 7.26 and Illegal Sources ¼ 8.34), a square root transformation was performed in each case to achieve normality. 3.3.2. Correlates of income generation Three categories of correlates were included in the model: demographic variables, homelessness stressors, and coping strategies. 3.3.2.1. Demographic variables. Demographic information included Age; Gender (0 ¼ female, 1 ¼ male); Race/Ethnicity (1 ¼ white, 2 ¼ black, 3 ¼ Hispanic, 4 ¼ other); and Education (0 ¼ dropped out, suspended, or still enrolled; 1 ¼ high-school degree or General Educational Development [GED]). Race/ethnicity was subsequently recoded using two dummy variables for the predominant racial/ ethnic groups in the sample (0 ¼ other, 1 ¼ White; and 0 ¼ other, 1 ¼ Black). To control for inter-city differences, the city in which data were collected was coded (1 ¼ Los Angeles, 2 ¼ Denver, 3 ¼ Austin) and then dummy-coded as Los Angeles (0 ¼ no, 1 ¼ yes) and Austin (0 ¼ no, 1 ¼ yes) with Denver as a reference category. 3.3.2.2. Homelessness stressors. Length of Housing Instability was determined by subtracting the number of months since the young adults had left home for good from the interview date. Arrest History assessed whether young adults had ever been arrested (0 ¼ no, 1 ¼ yes). Transience was measured as the total number of cities (new or repeated) to which young adults had moved since first leaving home. To assess for Peer Substance Use, young adults were asked how many of their friends over the past month had engaged in seven substance-use behaviors: gotten drunk, smoked marijuana, gotten high on inhalants, used cocaine, used heroin, used prescription drugs, or sold drugs (0 ¼ no friends, 1 ¼ some friends, 2 ¼ most friends). Responses were first dichotomized (0 ¼ no friends or some friends, 1 ¼ most friends). Responses of “1” (most friends use drugs) were then combined for each of the seven items into one interval variable (range 1e7). In addition, the Mini International Neuropsychiatric Interview (MINI, Version 6.0.0) was used to determine whether young adults met criteria for mental illness and substance use. The MINI is a widely used, brief, structured interview that screens for psychiatric disorders according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR). The MINI demonstrates good reliability as well as convergent validity with the Structured Clinical Interview for DSM-IV-TR Axis 1 Disorders (SCID), which is an established measure of diagnostic criteria (Sheehan et al., 1998). Antisocial Personality Disorder (ASPD) was determined by two or more affirmative responses to yes/no questions about behaviors committed before 15 years of age (e.g., repeatedly skipping school, lying, or stealing) and three or more affirmative responses to behaviors since 15 years of age (e.g., repeatedly fighting, exposing others to danger, or feeling no guilt after hurting others). Major Depressive Episode was determined by five or more affirmative responses to yes/no questions including: “Did you feel worthless or guilty almost every day?” Manic Episode was determined by three or more affirmative answers to a range of manic symptoms that lasted at least one week and caused significant problems at home, at work, socially or at school, or resulted in hospitalization. ASPD, major depressive episode, and manic episode were each dichotomous variables (0 ¼ no, 1 ¼ yes) measuring whether criteria were met. Lastly, Substance Use Disorder was determined through affirmative answers to screening

50

K.M. Ferguson et al. / Social Science & Medicine 135 (2015) 47e55

questions and a sufficient number of positive responses to symptom questions for both alcohol and drug dependence. This variable was measured as meeting the criteria for dependence of any drug or alcohol and was coded as 0 ¼ does not meet criteria, 1 ¼ meets criteria. 3.3.2.3. Coping strategies. Four dimensions of coping strategies were assessed using the Coping Scale (Kidd and Carroll, 2007). All coping items used a 5-point scale from 1 (Never) to 5 (Almost Always) in response to the prompt “Please rate how much you use each of the following ways of dealing with problems”. ProblemFocused Coping was assessed using two items: 1) “Concentrated on what to do and how to solve the problem” and 2) “Think about what happened and try to sort it out in my head”. Avoidant Coping was assessed with two items: 1) “Try not to think about it” and 2) “Go to sleep”. Social Coping was assessed with two items: 1) “Go to someone I trust for support” and 2) “Go off by myself to think,” indicating social withdrawal, both of which were derived from qualitative work by Kidd (2003). Lastly, Other Ways of Coping was comprised of eight additional coping strategies including “Use my anger to get me through it” and “Do a hobby (e.g., read, draw)” (Kidd, 2003). 3.4. Data analysis Researchers in each city first entered the raw interview data into a database created in SPSS (version 22.0) then combined the three city-level databases into one database. All predictor variables had less than 1.2% missing data, which were handled using the listwise deletion method in SPSS. There were no missing data on either outcome variable. Review of the variance inflation factors revealed that multicollinearity between independent variables would not significantly affect analyses; variance inflation factor statistics were below 2.16 (Austin dummy variable). Descriptive analyses were conducted to illustrate characteristics for the full sample and subsamples. Chi-square and independent t-tests were also used to identify gender differences. Six Ordinary Least Squares (OLS) regression models were created to analyze associations between correlates and legal and illegal sources of income among the full sample as well as separately with the female and male sub-samples. OLS regression analysis was first conducted with the full sample. Subsequently, separate models using the female and male sub-samples were conducted with the same correlates used with the full sample to examine gender differences. Two dummy variables reflecting three cities were included to control for city differences. All statistical tests were considered significant at p < .05. 4. Results 4.1. Sample characteristics Table 1 presents the characteristics of the full sample as well as female and male sub-samples. 4.2. Gender differences among income-generation sources and coping styles In the sub-sample of 216 females, 146 (67.6%) earned income via one or more legal sources (range ¼ 0e5) and 133 (61.6%) earned income through one or more illegal sources (range ¼ 0e4). Among the 385 males, 301 (78.2%) earned income via one or more legal sources (range ¼ 0e7) and 256 (66.5%) earned income through one or more illegal sources (range ¼ 0e5). Results of independent t-tests and chi-square analyses indicated

several gender differences in both the quantity and types of income-generating activities. Males earned income from a greater variety of both legal and illegal sources than did females. Males were also more likely than females to earn income from four economic activities: temporary work (45.2% vs. 26.2%, p < .001), dealing drugs (25.2% vs. 16.2%, p < .05), stealing (26.8% vs. 19.0%, p < .05), and gambling (14.0% vs. 5.1%, p < .01). Independent t-tests further indicate gender differences in coping styles. Males reported higher problem-focused coping scores than females (7.83 vs. 7.47, p < .05). In contrast, females had higher avoidant coping scores (6.19 vs. 5.51, p < .001) and higher social coping scores (7.45 vs. 7.05, p < .01) compared with males. 4.3. Correlates of source of income generation in multivariate models Table 2 presents the two OLS regression models for the full sample identifying relationships between income generation source and the combination of independent variables. 4.3.1. Legal income among full sample The first column in Table 2 displays the full model of determinants that was significantly associated with homeless young adults' legal income generation (F ¼ 2.06 [19], p < .01). Homeless young adults who relied on fewer avoidant coping strategies reported earning more diverse legal sources of income than their peers who relied on greater avoidant coping strategies. Although only marginally significant, those using more problem-focused coping strategies reported earning income from a greater variety of legal sources than those who relied on fewer problem-focused coping strategies. Also, homeless young adults who did not meet the criteria for manic episode reported a greater variety of legal sources of income than their peers who met the criteria. 4.3.2. Illegal income among full sample The second column in Table 2 displays the full model that was significantly associated with homeless young adults' illegal income generation (F ¼ 11.43 [19], p < .001). Homeless young adults who relied on fewer social coping strategies reported a greater variety of illegal sources than their peers who relied on more social coping strategies. Those who were younger reported a greater variety of illegal income sources than their older peers. Those who had previously been arrested reported more diverse illegal sources of income than those without an arrest history. Homeless young adults with greater transience and substance-using peers also reported more diverse illegal income sources than those with fewer intercity moves and substance-using peers respectively. Lastly, those who met the criteria for ASPD, substance use disorder, and major depressive episode reported a greater variety of illegal sources of income than their peers who did not meet criteria for these diagnoses. Table 3 presents the four OLS regression models for the female and male sub-samples. 4.3.3. Legal income among females The first column in Table 3 displays the full model that was significantly associated with females' legal income generation (F ¼ 2.44 [18], p < .01). Females who used greater problem-focused coping strategies reported earning more diverse legal sources of income than their peers who used fewer of these coping strategies. Those who were of Black, Hispanic, or other race/ethnicity earned income from a greater variety of legal sources than their White peers. Although only marginally significant, females with greater transience reported income from more diverse legal sources than their less transient counterparts.

K.M. Ferguson et al. / Social Science & Medicine 135 (2015) 47e55

51

Table 1 Characteristics of homeless young adults: full sample and sub-sample by gender. Variables

Gender Male Female Race/ethnicitya White Black Latino Other Education High-school degree/GED City located Los Angeles Austin Denver Homelessness stressorsb Ever arrested Antisocial personality disorder Substance use disorder Major depressive episode Manic episode Income generationb Legal sources Full-time employment Part-time employment Temporary paid employment Selling self-made items Selling bottles/cans Selling clothing/personal possessions Selling blood/plasma Illegal sources Panhandling Dealing drugs Survival sex Stealing Gambling Age (years) Homelessness stressorsc Length of housing instability (months) Transience Peer substance use Coping strategiesd Problem-focused Avoidant Social Other ways of coping Income generatione Legal sources Illegal sources

Full sample (N ¼ 601)

Females (n ¼ 216)

Males (n ¼ 385) n

%

385

100.0

Chi-square/t statistic

N

%

n

%

385 216

64.1 35.9

216

100.0

240 152 107 101

39.9 25.3 17.8 16.8

78 62 38 38

36.1 28.7 17.6 17.6

162 90 69 63

42.1 23.4 17.9 16.4

286

47.6

102

47.2

184

47.8

200 200 201

33.3 33.3 33.4

72 72 72

33.3 33.3 33.3

128 128 129

33.2 33.3 33.5

419 209 292 187 136

69.7 34.8 48.6 31.1 22.6

126 73 100 76 57

58.3 33.8 46.3 35.2 26.4

293 136 192 111 79

76.1 35.3 49.9 28.8 20.5

108 189 231 113 98 172 61

18.0 31.4 38.4 18.8 16.3 28.6 10.1

31 61 57 36 37 61 18

14.4 28.2 26.4 16.7 17.1 28.2 8.3

77 128 174 77 61 111 43

20.0 33.2 45.2 20.0 15.8 28.8 11.2

300 132 35 144 65 Mean

49.9 22.0 5.8 24.0 10.8 SD

108 35 14 41 11 Mean

50.0 16.2 6.5 19.0 5.1 SD

192 97 21 103 54 Mean

49.9 25.2 5.5 26.8 14.0 SD

20.1

1.6

19.9

1.6

20.1

1.6

32.4 3.5 2.6

31.0 3.7 1.5

29.2 3.2 2.6

28.1 3.5 1.5

34.3 3.7 2.6

32.5 3.8 1.4

t ¼ 2.0*

7.7 5.8 7.2 27.1

1.9 2.1 1.6 4.7

7.5 6.2 7.5 26.8

1.9 2.1 1.5 4.8

7.8 5.5 7.1 27.2

1.8 2.0 1.7 4.6

t ¼ 2.3* t ¼ 3.9*** t ¼ 2.9**

1.6 1.1

1.4 1.1

1.4 1.0

1.3 1.0

1.7 1.2

1.4 1.2

t ¼ 2.9** t ¼ 2.7**

c2 ¼ 20.7***

c2 ¼ 20.68***

c2 ¼ 6.53* c2 ¼ 4.59* c2 ¼ 11.45** t statistic

Note. *p < .05, **p < .01, ***p < .001. a Other race/ethnicity includes Asian, American Indian, mixed race, and other race/ethnicity. b Responses are dichotomized (0 ¼ no, 1 ¼ yes). c Transience represents the total number of cities (new or repeated) the young adults had moved to since first leaving home. Peer substance abuse is represented by an interval variable, ranging from 1 to 7, that captures how many of their friends engaged in seven substance-use behaviors during the last month: gotten drunk, smoked marijuana, gotten high on inhalants, used cocaine, used heroin, used prescription drugs, or sold drugs. Responses were first dichotomized (0 ¼ no friends or some friends, 1 ¼ most friends). Responses of 1 ¼ most friends use drugs were then combined for each of the seven items. d Four dimensions of coping strategies were assessed using the “Coping Scale” (Kidd and Carroll, 2007), which uses a 5-point scale from 1 (Never) to 5 (Almost Always); “Other Ways of Coping” included eight additional coping strategies: trying to learn from the bad experience, using my anger to get through it, using drugs or alcohol, doing a hobby, trying to value myself and not thinking so much about other people's opinions, realizing that I am strong and can deal with whatever is bothering me, thinking about how things will get better in the future, and using my spiritual beliefs/belief in a higher power (Kidd, 2003). e Legal and illegal sources are each represented by summary scores that capture all of the legal and illegal income sources for young people over the last 6 months; scores range from 0 to 7 for legal sources; 0e5 for illegal sources.

4.3.4. Illegal income among females The second column in Table 3 displays the full model that was significantly associated with females' illegal income generation (F ¼ 4.98 [18], p < .001). Females with higher problem-focused coping strategies reported a greater variety of illegal sources than their peers who employed fewer problem-focused coping

strategies. Although only marginally significant, females who reported fewer social coping strategies also earned income from a greater variety of illegal sources than their peers who used more of these coping strategies. Females who had been arrested reported more diverse illegal sources of income than those without an arrest history. Females with a greater number of substance-using peers

52

K.M. Ferguson et al. / Social Science & Medicine 135 (2015) 47e55

Table 2 Regression models of legal and illegal income generation for full sample. Variables

Model 1: full sample legal

Demographics Gender Age (years) White dummy Black dummy High school/GED Los Angeles dummy Austin dummy Homelessness stressors Length of housing instability Arrest history Transience Peer substance use Antisocial personality disorder Substance use disorder Major depressive episode Manic episode Coping strategies Problem-focused coping Avoidant coping Social coping Other ways of copingc R-square

Model 2: full sample illegal

B (b)a

SEb

t

B (b)a

SEb

t

.10 (.07) .02 (.04) .04 (.03) .09 (.05) .06 (.04) .06 (.04) .08 (.05)

.07 .02 .09 .09 .07 .09 .10

1.33 .69 .41 .91 .85 .64 .76

.07 (.05) .06 (.14)** .05 (.04) .09 (.06) .01 (.01) .00 (.00) .23 (.18)**

.06 .02 .07 .07 .05 .07 .08

1.23 3.10 .71 1.22 .13 .00 3.01

.00 (.02) .13 (.08) .06 (.09) .01 (.03) .04 (.03) .10 (.07) .10 (.07) .14 (.09)y

.00 .08 .04 .02 .07 .08 .08 .08

.31 1.55 1.52 .54 .53 1.29 1.26 1.74

.00 (.03) .15 (.10)* .06 (.10)* .09 (.21)*** .29 (.22)*** .16 (.13)** .13 (.09)* .08 (.05)

.00 .07 .03 .02 .06 .06 .06 .06

.62 2.31 2.07 5.05 5.10 2.79 2.10 1.23

.04 (.10)y .04 (.11)* .00 (.00) .00 (.03) .084

.02 .02 .02 .01

1.80 2.09 .07 .48

.02 (.05) .01 (.02) .05 (.14)** .01 (.05) .338

.02 .01 .02 .01

.98 .38 3.09 1.14

y

p < .10; *p < .05; **p < .01; ***p < .001. a Unstandardized regression coefficient (B) is followed by standardized regression coefficient (Beta [b]) in parentheses. b SE ¼ Standard error. c “Other Ways of Coping” included eight additional coping strategies: trying to learn from the bad experience, using my anger to get through it, using drugs or alcohol, doing a hobby, trying to value myself and not thinking so much about other people's opinions, realizing that I am strong and can deal with whatever is bothering me, thinking about how things will get better in the future, and using my spiritual beliefs/belief in a higher power (Kidd, 2003).

Table 3 Regression models of legal and illegal income generation for female and male sub-samples. Variables

Demographics Gender Age (years) White dummy Black dummy High school/GED Los Angeles dummy Austin dummy Homelessness stressors Length of housing instability Arrest history Transience Peer substance use Antisocial personality disorder Substance use disorder Major depressive episode Manic episode Coping strategies Problem-focused coping Avoidant coping Social coping Other ways of copingc R-square

Model 3: females legal

Model 5: males legald

Model 4: females illegal

Model 6: males illegal

B (b)a

SEb

t

B (b)a

SEb

t

B (b)a

SEb

t

B (b)a

SEb

t

.02 (.05) .31 (.21)* .23 (.14) .16 (.11) .07 (.04) .29 (.19)

.04 .15 .16 .12 .16 .18

.57 2.02 1.49 1.37 .47 1.66

.05 (.14)y .06 (.05) .17 (.12) .07 (.05) .00 (.00) .28 (.22)*

.03 .12 .12 .09 .12 .13

1.71 .51 1.46 .73 .02 2.07

.05 (.11) .07 (.05) .04 (.02) .01 (.01) .00 (.00) .10 (.07)

.03 .11 .12 .09 .12 .12

1.54 .60 .33 .13 .02 .77

.06 (.15)* .02 (.01) .06 (.04) .03 (.02) .03 (.02) .21 (.16)*

.02 .09 .10 .07 .10 .10

2.58 .16 .59 .36 .36 2.11

.00 (.03) .15 (.10) .11 (.17)y .01 (.01) .01 (.01) .07 (.05) .05 (.04) .02 (.01)

.00 .12 .06 .04 .12 .13 .14 .14

.26 1.25 1.67 .13 .09 .55 .39 .13

.00 .20 .01 .08 .32 .14 .06 .00

(.01) (.16)* (.02) (.19)* (.25)** (.11) (.05) (.00)

.00 .09 .05 .03 .09 .10 .10 .10

.08 2.20 .19 2.62 3.44 1.42 .61 .03

.00 (.03) .06 (.03) .02 (.03) .00 (.01) .08 (.05) .09 (.06) .11 (.07) .24 (.14)

.00 .12 .04 .03 .09 .09 .10 .11

.39 .46 .37 .10 .82 .94 1.13 2.19

.00 (.06) .07 (.05) .07 (.13)* .09 (.20)*** .28 (.21)*** .17 (.13)* .16 (.11)* .14 (.09)

.00 .10 .03 .03 .07 .08 .08 .09

.99 .78 2.11 3.75 3.82 2.28 2.04 1.56

.12 (.31)** .04 (.11) .04 (.08) .02 (.11) .238

.03 .03 .04 .02

3.51 1.36 .95 1.11

.05 (.16)* .02 (.06) .05 (.13)y .00 (.00) .389

.03 .02 .03 .01

1.99 .85 1.67 .05

.01 (.03) .03 (.10) .03 (.06) .01 (.07) .062

.03 .02 .03 .01

.42 1.47 .92 .10

-.01 (-.01) .01 (.02) .05 (.13)* .01 (.06) .326

.02 .02 .02 .01

.23 .29 2.33 1.05

yp < .10; *p < .05; **p < .01; ***p < .001. a Unstandardized regression coefficient (B) is followed by standardized regression coefficient (Beta [b]) in parentheses. b SE ¼ Standard error. c “Other Ways of Coping” included eight additional coping strategies: trying to learn from the bad experience, using my anger to get through it, using drugs or alcohol, doing a hobby, trying to value myself and not thinking so much about other people's opinions, realizing that I am strong and can deal with whatever is bothering me, thinking about how things will get better in the future, and using my spiritual beliefs/belief in a higher power (Kidd, 2003). d Model not significant for legal income sources with male sub-sample.

K.M. Ferguson et al. / Social Science & Medicine 135 (2015) 47e55

also reported more diverse illegal sources than those with fewer substance-using peers. Females who met the criteria for ASPD reported more diverse illegal sources of income than their peers who did not meet criteria for this diagnosis. Lastly, although only marginally significant, younger females reported a greater variety of illegal sources of income than their older peers. 4.3.5. Legal income among males The model depicting legal income generation among the male sub-sample was not significant yet parameters are listed in the third column of Table 3. 4.3.6. Illegal income among males The fourth column in Table 3 displays the full model that was significantly associated with males' illegal income generation (F ¼ 7.17 [18], p < .001). Males who relied on fewer social coping strategies reported a greater variety of illegal income sources than their peers who relied on more of these coping strategies. Younger males reported a greater variety of illegal sources of income than their older peers. Males with greater transience and substanceusing peers also reported more diverse illegal sources of income than their peers with fewer inter-city moves and substance-using peers. Lastly, those who met the criteria for ASPD, substance use disorder, and major depressive episode each reported income from a greater variety of illegal sources than males who did not meet diagnostic criteria. 5. Discussion and implications Findings provide a greater understanding of the risk and resilience factors associated with legal and illegal income generation among homeless young adults and, in particular, how these factors differ by gender. Several findings from this study are important to highlight. First, specific coping strategies such as problem-focused coping, functioned as protective factors, buffering youth from the effects of well-established risk factors among homeless young people (e.g., criminal behavior, transience, mental illness, and substance use). That homeless young people relied on positive coping techniques is important to note, given that coping skills are often learned in the context of family upbringing (Kidd, 2003). In the case of these homeless young people, many of whom originated from dysfunctional family environments, they were able to learn prosocial coping strategies from other contexts (e.g., supportive adults and institutions, peer groups, etc.). In this study, homeless young adults who earned income from legal sources tended to rely on greater problem-focused and fewer avoidant coping strategies. In particular, that problem-focused coping (i.e., efforts to address the stressor itself) was associated with legal income generation suggests this coping style is protective for homeless young people and might facilitate their pursuit of safe and legal economic activity. Prior research with persons experiencing mental illness suggests that higher cognitive functioning and problem-solving are among the characteristics associated with successful formal employment outcomes (Campbell et al., 2010). Clinicians' efforts to strengthen homeless young people's cognitive functioning and problemsolving abilities using cognitive-behavioral techniques such as cognitive restructuring might, in turn, equip these young adults with coping strategies that aid them in formal employment settings (Hope et al., 2010) as well as in navigating and exiting homelessness (Kidd, 2003). Likewise, homeless young people who displayed fewer avoidant coping strategies (i.e., problem and emotional avoidance as well as sleep) also experienced greater involvement in legal employment. Through employment, individuals benefit from time structure, a sense of responsibility, and social contactsdall of which positively

53

influence their well-being (Harnois and Gabriel, 2000). It might be that safe and legal economic activity enhances these young people's personal and professional responsibilities including actively addressing stressors that could negatively impact their work performance. This finding suggests that efforts to support homeless young people in identifying their stressors and devising strategies to address them might help this population to pursue formal employment and safe, legal informal work. Clinical interventions drawing on social cognitive theory that strengthen these young people's self-efficacy and belief in their abilities to succeed (Bandura, 1994) might provide them with the skills needed to thrive in formal employment and exit homelessness. Among persons with mental illness, greater self-efficacy is associated with successful employment outcomes (Campbell et al., 2010). Yet homeless young people commonly experience psychological challenges related to repeated exposure to trauma and loss, both within their families and once homeless, as well as significant identity disruption during their transition to adulthood (Bender et al., 2014; Whitbeck, 2009). Both complex trauma and identity disruption affect coping skills, which can compromise successful employment outcomes in this population (Kidd, 2003). As such, additional interventions for homeless young people are needed to supplement traditional cognitive restructuring and coping skillstrainings in order to improve youth's psychological functioning and support them in their transition to adulthood. As a complement to their employment services, homeless young adults also could benefit from clinical services to help them address trauma, loss, and identity disruption, including securing safe and supportive housing, preventing victimization and re-victimization once homeless, building trust with service providers and other adult role models, and decreasing harmful coping techniques such as substance use. With respect to coping as a risk factor, homeless young people who relied less often on social coping strategies were more likely to engage in illegal activity in the full sample and male and female sub-samples. This population often experiences social estrangement and isolation in particular the longer they are homeless (Baron, 2004). Prolonged social estrangement and isolation can, in turn, affect the prosocial coping strategies that are characteristic of homeless young people engaged in formal employment. One explanation for the association between social isolation and illegal economic activity is that the disconnection from supportive adults and institutions might preclude these young people from access to formal employment opportunities and settings. Due to their labor exclusion, they often turn to homeless peers for emotional and instrumental support (Barman-Adhikari and Rice, 2014) as well as illegal activities to meet their basic needs (O'Grady and Gaetz, 2004). This finding highlights the need to connect homeless young people with supportive adults and institutions as well as pro-social peers to enhance their employment outcomes. In one study, homeless youth who received emotional support from their homeless peers were more likely to engage in agency-based employment services than their peers without such support (Barman-Adhikari and Rice, 2014). Connecting homeless youth to agency-based employment services with supportive staff as well as job training and placement is an important first step in exposing them to caring adults who can model problem-solving and support them in their long-term employment goals, which commonly include aspirations of formal employment (O'Grady and Gaetz, 2004). Several gendered coping differences were detected that further illustrate how distinct coping strategies function as protective and/ or risk factors. For instance, a greater use of problem-focused coping strategies was associated with legal and illegal income generation in females. There was no association found between

54

K.M. Ferguson et al. / Social Science & Medicine 135 (2015) 47e55

problem-focused coping and legal or illegal income generation in males. It might be that high cognitive functioning, self-reliance, and problem solving (characteristic of problem-focused coping) not only help females manage homelessness stressors but also constitute assets in legal (formal) employment as well as in informal (illicit) work. That the streets have been defined as a “male” space in which males hold greater power (O'Grady and Gaetz, 2004) might require homeless females to learn to solve problems rapidly and exhibit self-reliance to successfully navigate this space and the illicit economic opportunities the streets afford them. Thus, adopting problem-solving coping strategies might enable these young women to thrive in legal employment and in the streets. Other studies examining gender differences in coping styles with housed youth found that females scored higher than males in problem-solving coping (Eschenbeck et al., 2007). Yet with respect to coping among homeless young adult females, problem-focused coping functioned as both a protective factor for legal employment as well as a risk factor for illegal employment. This finding suggests that the risk and resilience framework might be overly simplistic for understanding the coping strategies of homeless young adults. Their involvement in varied conventional and unconventional economic activities can be considered both protective and opportunistic, since they often need to take responsibility for themselves, including achieving financial independence from their families of origin, at an earlier age than their housed counterparts (Whitbeck, 2009). Another important finding was that transience had a differential impact on economic activity depending on gender. Greater transience was associated with more diverse involvement in legal activities in females (albeit at the p < .10 level), whereas transience was associated with more diverse illegal activities in males. It might be that for homeless females, transience offers an opportunity to exit homelessness (e.g., through a job or housing opportunity in another city); yet for males, transience is a way of becoming more entrenched in homelessness. If this is the case, males might experience greater obstacles to exiting homelessness than females since the longer they engage in street activity, the harder it becomes for them to transition off the streets and into formal employment and housing, which often require major changes in their attitudes and behaviors (Whitbeck, 2009). With these findings in mind, when transience is associated with formal employment and safe, legal informal work, service providers can support homeless young people (in particular females) in making travel safer and mobile services more accessible (e.g., via technology-enhanced services). Conversely, when transience is associated with illegal activity, providers can help strengthen these youth's (in particular males') problem- and social-focused coping strategies to overcome adversity and to prevent the use of transience as a coping mechanism. Other interesting gender differences emerged regarding the young adults' mental health backgrounds. For example, both depression and substance use disorder were associated with illegal income generation in males yet not in females. Mental illness and substance use are oft-cited barriers to formal employment (Bond and Drake, 2008) as well as facilitators of illicit activity (Baron, 2004). The fact that males were more likely than females to earn income from three types of illegal activities (i.e., dealing drugs, stealing, and gambling) suggests that perhaps their mental health symptoms and substance use hindered their involvement in fulland part-time formal employment and instead contributed to their involvement in illicit economic activity (or in legal temporary work). It might be that securing short-term work through temp agencies or day-labor sites is a more common activity among males (Robinson and Baron, 2007). It might also be that temporary work is more attractive to male homeless young people, who may be able to cope with their symptoms of depression and substance use on a

temporary basis, yet not over the long-term as would be required for ongoing full- and part-time work. With respect to legal correlates of economic activity, it is noteworthy that prior arrest history was associated with illegal economic activity in females yet not in males. One explanation for this gendered finding involves these young women's meeting criteria for ASPD, which was also associated with illegal economic activity in females. Research with homeless adults reveals that the combination of an ASPD diagnosis and prior arrest history was significantly associated with recurrent homelessness (McQuistion et al., 2014). It might be that homeless female youth's symptoms of ASPD, such as committing illegal activities, ultimately result in arrests and incarceration, hinder formal employment, and lure these young women into illicit economic activity for survival (Baron, 2004; Whitbeck, 2009). It also might be that homeless females who have experienced prior sexual abuse might also experience ASPD symptoms, which place them at greater risk for involvement in sex work in the public space of the streets where they are at greater risk of being arrested (Cale and Lilienfeld, 2002; O'Grady and Gaetz, 2004). Careful consideration should be given to the context within which the ASPD symptoms measured here evolved in this population. Such traits as conning others, stealing, or intimidating others, while technically aligning with DSM criteria for ASPD, might also represent self-protection and survival strategies particularly important for females to navigate street life. Future research is also needed to better understand how homeless females' arrest histories and the street milieu affect their ASPD symptoms and their decisions to engage in illegal activity. 5.1. Limitations Certain limitations influence the interpretation of study findings. First, although the data collection was similar across cities, the samples of service-seeking homeless young adults might not be representative of the population in any of the cities given the use of non-probability sampling. These findings thus might not generalize to other non-service-using homeless youth populations including traveling homeless young people or those with small children who instead might access family shelters. Similarly, the choice of cities and host agencies was based on feasibility, not on representativeness of homeless young adults. Additionally, the cross-sectional design and use of individual (versus structural) correlates limited the ability to draw conclusions on homeless young adults' income generation over time. It is not possible to determine the positive and negative outcomes associated with involvement in legal versus illegal work since data were collected at only one time point. Similarly, all variables measured individual characteristics of homeless youth. Indeed, understanding and addressing homelessness (and the role of formal employment in addressing homelessness) requires attention to the structural roots contributing to its existence such as poverty, unemployment, lack of affordable housing, and discrimination (Hopper, 2003). Future longitudinal studies that include both individual and structural correlates of homelessness and use multiple follow-ups over several years can begin to address these challenges. Further, the use of illegal behavior variables is a concern in studies using self-report data. Although the interviewers had prior research and practice experience with homeless youth, it is possible that the oral interviewing format introduced bias into this study. Participants might have under-reported (or over-reported) their responses about stigmatized risk behaviors (e.g., survival sex, drug use, criminal activity) due to social desirability. Lastly, although our overall measure of psychiatric diagnostic criteria (i.e., MINI) has been well validated and found to be reliable,

K.M. Ferguson et al. / Social Science & Medicine 135 (2015) 47e55

there is little available validation of the ASPD module specifically within the empirical literature. As the ASPD variable was predictive of outcomes in this study, the conceptualization of ASPD should be interpreted with caution. 6. Conclusions Homeless male and female young adults experience different risk factors related to their homeless experience, utilize different strategies to cope with risks, and engage in different types of economic activity. As such, practitioners and policy makers would benefit from customizing employment and clinical interventions to reduce risk associated with illicit economic activity and reinvest homeless youth's entrepreneurial skills in safe and legal income generation. Further, investing in clinical services and trainings to aid this population in maximizing problem-focused and minimizing avoidant coping strategies will equip them with needed skills to succeed in formal employment. Finally, a greater understanding of the ways in which male and female homeless young adults experience and cope with adversity might also guide the development of customized prevention and intervention efforts aimed at safe and legal income generation. Acknowledgments Funding for this study was provided in Los Angeles by the University of Southern California (USC), School of Social Work Hamovitch Research Center; in Denver by the University of Denver, Graduate School of Social Work; and in Austin by a Faculty Development Grant from the University of Texas at Austin and the Center for Social Work Research. We would like to acknowledge Connie Chung from the Graduate School of Education at Harvard University, Kimberly Biddle and Jina Sang from the USC School of Social Work, Jamie Yoder and Chelsea Komlo from the University of Denver, and Tiffany Ryan, Katherine Montgomery, and Angie Lippman from the University of Texas at Austin for their involvement in the study as research assistants. References Bandura, A., 1994. Self-efficacy. In: Ramachaudran, V.S. (Ed.), Encyclopedia of Human Behavior, vol. 4. Academic Press, New York, pp. 71e81. Barman-Adhikari, A., Rice, E., 2014. Social networks as the context for understanding employment services utilization among homeless youth. Eval. Program Plan. 45, 90e101. http://dx.doi.org/10.1016/j.evalprogplan.2014.03.005. Baron, S.W., 2004. General strain, street youth and crime: a test of Agnew's revised theory. Criminology 42 (2), 457e483. http://dx.doi.org/10.1111/j.17459125.2004.tb00526.x. Baron, S.W., Hartnagel, T.F., 1997. Attributions, affect, and crime: street youths' reactions to unemployment. Criminology 35 (3), 409e434. http://dx.doi.org/ 10.1111/j.1745-9125.1997.tb01223.x. Bender, K., Thompson, S.J., Ferguson, K., Yoder, J., Kern, L., 2014. Trauma among street- involved young adults. J. Emot. Behav. Disord. 22 (1), 53e64. http:// dx.doi.org/10.1177/1063426613476093. Bond, G.R., Drake, R.E., 2008. Predictors of competitive employment among patients with schizophrenia. Curr. Opin. Psychiatr. 21 (4), 362e369. http://dx.doi.org/10. 1097/YCO.0b013e328300eb0e. Cale, E.M., Lilienfeld, S.O., 2002. Sex differences in psychopathy and antisocial

55

personality disorder: a review and integration. Clin. Psychol. Rev. 22 (8), 1179e1207. http://dx.doi.org/10.1016/S0272-7358(01)00125-8. Campbell, K., Bond, G.R., Drake, R.E., McHugo, G.J., Xie, H., 2010. Client predictors of employment outcomes in high-fidelity supported employment: a regression analysis. J. Nerv. Ment. Dis. 198 (8), 556e563. http://dx.doi.org/10.1097/NMD. 0b013e3181ea1e53. Dachner, N., Tarasuk, V., 2002. Homeless “squeegee kids”: food insecurity and daily survival. Soc. Sci. Med. 54, 1039e1049. http://dx.doi.org/10.1016/S02779536(01)00079-X. Eschenbeck, H., Kohlmann, C., Lohaus, A., 2007. Gender differences in coping strategies in children and adolescents. J. Individ. Differ. 28 (1), 18e26. http:// dx.doi.org/10.1027/1614-0001.28.1.18. Ferguson, K., Bender, K., Thompson, S., Xie, B., Pollio, D., 2011. Correlates of streetsurvival behaviors in homeless young adults in four U.S. cities. Am. J. Orthopsychiatr. 81 (3), 401e409. http://dx.doi.org/10.1111/j.1939-0025.2011.01108.x. Ferguson, K., Xie, B., Glynn, S., 2012. Adapting the individual placement and support model with homeless young adults. Child Youth Care Forum 41 (3), 277e294. http://dx.doi.org/10.1007/s10566-011-9163-5. Fraser, M.W., Galinsky, M.J., Richman, J.M., 1999. Risk, production, and resilience: toward a conceptual framework for social work practice. Soc. Work Res. 23 (3), 131e143. http://dx.doi.org/10.1093/swr/23.3.131. Gabriel, P.E., Schmitz, S., 2006. The impact of gender differences in occupational attainment on the relative earnings of young workers. Appl. Econ. Lett. 13 (10), 615e619. http://dx.doi.org/10.1080/13504850500401742. Gaetz, S., O'Grady, B., 2002. Making money: exploring the economy of young homeless workers. Work Employ. Soc. 16 (3), 433e456. http://dx.doi.org/ 10.1177/095001702762217425. Greene, J.M., Ennett, S.T., Ringwalt, C.L., 1999. Prevalence and correlates of survival sex among runaway and homeless youth. Am. J. Public Health 89 (9), 1406e1409. http://dx.doi.org/10.2105/AJPH.89.9.1406. Harnois, G., Gabriel, P., 2000. Mental Health and Work: Impact, Issues and Good Practices. World Health Organization and International Labor Organization. World Health Organization, Geneva. Hope, D.A., Burns, J.A., Hayes, S.A., Herbert, J.D., Warner, M.D., 2010. Automatic thoughts and cognitive restructuring in cognitive behavioral group therapy for social anxiety disorder. Cognitive Ther. Res. 34 (1e12) http://dx.doi.org/10.1007/ s10608-007-9147-9. Hopper, K., 2003. Reckoning with Homelessness. Cornell University Press, New York. Kidd, S.A., 2003. Street youth: coping and interventions. Child Adolesc. Soc. Work J. 20 (4), 235e261. Kidd, S.A., Carroll, M.R., 2007. Coping and suicidality among homeless youth. J. Adolesc. 30 (2), 283e296. http://dx.doi.org/10.1016/j.adolescence.20 06.03.002. McQuistion, H.L., Gorroochurn, P., Hsu, E., Caton, C.L., 2014. Risk factors associated with recurrent homelessness after a first homeless episode. Community Ment. Health J. 50 (5), 505e513. http://dx.doi.org/10.1007/s10597-013-9608-4. O'Grady, B., Gaetz, S., 2004. Homelessness, gender and subsistence: the case of Toronto street youth. J. Youth Stud. 7 (4), 397e416. http://dx.doi.org/10.1080/ 1367626042000315194. Rew, L., Taylor-Seehafer, M., Thomas, N.Y., Yockey, R.D., 2001. Correlates of resilience in homeless adolescents. J. Nurs. Scholarsh. 33 (1), 33e40. http://dx.doi.org/ 10.1111/j.1547-5069.2001.00033.x. Robinson, J.L., Baron, S.W., 2007. Employment training for street youth: a viable option? Can. J. Urban Res. 16 (1), 33e57. Sheehan, D.V., Lecrubier, Y., Sheehan, K.H., Amorim, P., Janavs, J., Weiller, E., Dunbar, G.C., et al., 1998. The Mini International Neuropsychiatric Interview (MINI): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. J. Clin. Psychiatr. 59, 22e33. Silver, E., 2000. Extending social disorganization theory: a multilevel approach to the study of violence among persons with mental illness. Criminology 38 (4), 1043e1074. http://dx.doi.org/10.1111/j.1745-9125.2000.tb01414.x. Unger, J.B., Kipke, M.D., Simon, T.R., Johnson, C.J., Montgomery, S.B., Iverson, E., 1998. Stress, coping, and social support among homeless youth. J. Adolesc. Res. 13, 134e157. http://dx.doi.org/10.1177/0743554898132003. Votta, E., Manion, I.G., 2003. Factors in the psychological adjustment of homeless adolescent males: the role of coping style. J. Am. Acad. Child Adolesc. Psychiatr. 42, 778e785. http://dx.doi.org/10.1097/01.CHI.0000046871.56865.D9. Whitbeck, L.B., 2009. Mental Health and Emerging Adulthood among Homeless Young People. Psychology Press, New York.