Table of Contents
General Strain Predictors of Arrest History Among Homeless Youths from Four United States Cities
Kristin M. Ferguson
Hunter College, City University of New York
University of Denver, Denver, Colorado
Sanna J. Thompson
University of Texas at Austin
Claremont Graduate University, Claremont, California
University of Alabama, Tuscaloosa
Kristin M. Ferguson, Silberman School of Social Work at Hunter College, City University of New York; Kimberly Bender, School of Social Work, University of Denver, Denver, Colorado; Sanna J. Thompson, School of Social Work, University of Texas at Austin; Bin Xie, School of Community and Global Health, Claremont Graduate University, Claremont, California; and David Pollio, School of Social Work, University of Alabama, Tuscaloosa
Correspondence concerning this article should be addressed to: Kristin M. Ferguson, Silberman School of Social Work at Hunter College, 2180 Third Avenue, New York, NY 10035; E-mail: email@example.com
Funding for this study was provided in Los Angeles by the Haynes Foundation; in Denver by the University of Denver, Graduate School of Social Work; in New Orleans by Louisiana State University’s Council on Faculty Research Grant (FRG) program; and in St. Louis by the Center for Mental Health Services Research, George Warren Brown School of Social Work, Washington University in St. Louis.
We would like to acknowledge Gretchen Heidemann, MSW, from the University of Southern California; Jennifer McClendon, PhD, from Adelphi University; Daniela Young, from the University of Denver; and Elaine M. Maccio, PhD, Louisiana State University, for assistance with data collection, entry, and analysis.
Keywords: Homeless youth, crime, arrest, path model, general strain theory
This study identifies mental health and situational predictors of arrest history among homeless youth in four U.S. cities. Using convenience sampling, we recruited 188 homeless youths from shelters, drop-in centers, and street outreach using similar methods. The youths, aged 18–24, came from Los Angeles, California (n = 50), Denver, Colorado (n = 50), New Orleans, Louisiana (n = 50), and St. Louis, Missouri (n = 38). General strain theory provided a framework for identifying factors related to arrest history, including length of time homeless, level of transience, victimization, post-traumatic stress disorder (PTSD), substance dependence, and the use of survival strategies. We tested the general strain model using observed-variable path analysis. Collectively, youths’ length of time homeless, drug dependence, and use of survival strategies explained 17% of the variance in arrest history. We found a significant overall mediation effect from transience to arrest history through greater victimization, post-traumatic stress disorder, drug dependence, and survival strategies. This study offers one of the first applications of general strain theory to identify both mental health and situational strains—and responses to strains—among homeless youth. Findings have important implications for research and preventive interventions to address delinquency among this population.
More than 2 million youth experience homelessness in the United States each year (Whitbeck, 2009). They may include runaway-homeless youth, who have left home for one or more nights without notifying their parents or guardians; throwaway youth, who have left home because their parents have asked them to leave or have locked them out; or independent youth, who do not have a home to which they can return. Youth who live on the streets may also be part of biological homeless families or fictive street families (Halcon & Lifson, 2004). Undocumented, unaccompanied youth, whose families often reside in the youths’ country of origin, also are part of the homeless youth population. Finally, emancipated foster youth, who have aged out of foster care, are disproportionately represented among homeless youth in many cities. This diverse group of homeless youth is at increased risk for committing delinquent behaviors—often in reaction to environmental stressors or out of necessity for survival (Gaetz & O’Grady, 2002). Thus, the homeless youth population overlaps with more general delinquent youth populations, but is unique in that youths’ experiences of homelessness are intertwined with their engagement in delinquent behaviors. Differentiating the unique predictors of homeless youths’ delinquent behaviors will inform crime prevention efforts with this vulnerable population.
Prior research indicates that homeless youth are more likely than their housed peers to be involved in illegal activities, such as theft and property offenses, and drug possession, use, and sales (Baer, Peterson, & Wells, 2004; Thompson, Jun, Bender, Ferguson, & Pollio, 2010). Arrest rates for these young people range from 20%–30% (O’Grady & Gaetz, 2004). With annual estimates of 750,000 to 2 million homeless youth in the United States (Whitbeck, 2009), a conservative estimate translates to 150,000 of these young people encountering the criminal justice system in any given year.
Although research demonstrates that homeless youth engage in criminal activity, few studies have explored the complex interactions of risk factors associated with their arrest history. The present study goes beyond extant work (Baron 2004, 2008; Baron & Hartnagel, 1997; Whitbeck, Hoyt, & Yoder, 1999) in three ways. First, this study examines general strain theory predictors and mental health responses to such strains as they relate to arrest history among homeless youth. To date, most causal models of crime and delinquency draw from social learning and social control theories (Agnew, 1992). Second, this study is novel in testing the mediating effects of mental health and situational strain factors on homeless youths’ arrest history. Finally, rather than focusing on one or two cities or several cities within one region as most prior work has done, this study recruited a sample of homeless youth from four cities across disparate regions of the United States.
General Strain Theory and Homeless Youth
General strain theory posits that life strains and stressors result in negative emotional responses that may lead individuals to engage in criminal behaviors (Agnew, 1992). More specifically, this theory focuses on negative or inequitable relationships, such as parental abuse and neglect, which may influence individuals’ engagement in criminal behaviors (Baron, 2004). For homeless youth, viewing their homes of origin as highly inequitable environments may constitute pressures or strains that lead to emotional responses of anger and resentment (Whitbeck, 2009). Researchers suggest that the disorganized and abusive home environments of many homeless youth can engender anger and aggression in the youth as a reaction to their initial abusive relationships (Agnew, 1992; Baron 2004, 2008). It is likely that these strains contribute to the youth leaving home initially (Thompson, McManus, & Voss, 2006; Whitbeck, 2009).
General strain theory also posits that individuals have innate aspirations and expectations of achievement, and that the disparity between expectations and actual achievements can contribute to delinquent behavior (Agnew, 1992). The inability of individuals to achieve certain ideal goals that are emphasized by their societal or cultural systems (e.g., economic self-sufficiency) acts as a strain. As a result, deviant behaviors may become a possible option for achieving these goals or coping with the failure to achieve such goals. In the case of homeless youth, their low educational levels, combined with limited work histories, can hinder their success in obtaining and maintaining formal employment (Whitbeck, 2009). To meet their needs, many rely upon informal sources of income, both legal (e.g., panhandling and selling recycled/self-made items) and illegal (e.g., prostitution, theft, and selling drugs) (Gaetz & O’Grady, 2002; Kipke Unger, O’Connor, Palmer, & LaFrance, 1997).
The theory suggests that strains—and in particular chronic strains such as those experienced by familial abuse—exert pressure on individuals to engage in criminal behaviors. With respect to homeless youth, chronic stressors associated with homelessness include experiences of victimization, food and shelter insecurity, geographic mobility, and unemployment. These stressful experiences, combined with prior experiences of neglect and maltreatment, can lead many youths to use illegal substances and engage in antisocial behaviors in order to cope (Thompson, Maccio, Desselle, & Zittel-Palamara, 2007; Whitbeck et al., 1999).
Agnew (1992) suggested that some criminal behaviors may be understood as coping mechanisms. Strategies related to criminal behaviors include illegal drug use and violent behavior. These activities have been suggested as strategies that relieve or minimize the emotional severity of strains and provide a means of distraction and/or retaliation for the identified strain (Agnew, 1992; Baron, 2004). This understanding points to a chain of events in which strains lead to negative reactions (such as PTSD and other mental health challenges) and the development of coping strategies (such as substance use and survival strategies), which may ultimately culminate in arrest.
General strain theory provides a useful framework for examining strains associated with homeless youths’ arrest history. It provides a means of identifying specific strains that may lead these youths to interact and respond in antisocial, even criminal, ways. Although homeless youths experience a considerable amount of strain in their daily lives (Baron, 2004, 2008; Baron & Hartnagel, 1997), few studies have developed causal models of crime with variables derived from general strain theory to examine the interactions among stressors—and responses to these stressors—and how they collectively influence youths’ criminal arrest history (Agnew, 1992). To address this gap, this study examined the arrest histories of homeless youths in four U.S. cities in relation to various strains common among this population. These strains include length of time homeless, level of transience, and victimization. These strains may lead to reactions or responses, such as developing symptoms of PTSD and substance dependence as well as engaging in anti-social survival strategies often required to survive on the streets. Each of these strains and responses associated with such strains is discussed next.
Extended homelessness. The longer young people remain on the streets, the more they become entrenched in a street lifestyle characterized by inequitable and abusive relationships and interactions (Tyler & Johnson, 2006). Engaging in street life, combined with disengaging from traditional expectations (e.g., academic and employment achievement, monetary success), is associated with criminal behavior (Baron & Hartnagel, 1997). Increased exposure to and interactions with homeless peers facilitate acculturation to the streets and greater involvement in the street economy (Fest, 2003; Kipke et al., 1997). As a result, homeless youth who are embedded in abusive and inequitable relationships and who remain unstably housed may turn to criminal behaviors for economic survival or to cope with the daily stressors of a street lifestyle (Gaetz & O’Grady, 2002). Extended homelessness constitutes a strain in homeless youths’ lives as it influences their identities, needs, and goals while distancing them from expectations valued by traditional society (Baron 2004).
Transience. High levels of transience may be related to engaging in varying degrees of criminal activity, given that geographic mobility prohibits stable employment and housing (Ferguson, Bender, Thompson, Xie, & Pollio, 2011). Transient youth, by virtue of repeatedly moving from place to place, may be less likely than more stable youth to establish relationships with traditional institutions or to adopt traditional values. The lack of connections with trusted peers and adults—and negative interactions with street-involved individuals—may lead to an inability to provide for daily needs, resulting in engagement in the local street lifestyle to meet those needs (Bender, Thompson, McManus, Lantry, & Flynn, 2007). Constant relocation may also exacerbate strains associated with homelessness, including food insecurity, precarious housing, and hyper-vigilance concerning personal safety and belongings. Traveling homeless youth must locate safe places, supportive peers, and resources in each city (Dachner & Tarasuk, 2002). Thus, the strains associated with extended time on the streets and high transience are likely associated with engaging in illegal behaviors, though limited research has tested these relationships.
Maltreatment and victimization. Considerable evidence indicates that serious abuse occurs within families of youth who run away and become homeless (Whitbeck, 2009). Research suggests that once on the streets, those who remain for longer periods of time are at greater risk for victimization (Whitbeck, Hoyt, & Ackley, 1997). Homeless youth, especially females, are highly susceptible to victimization (Kushel, Yen, Gee, & Courtney, 2007), as they often live in precarious and dangerous situations. Strains from living on the street are commonplace among homeless youth, especially experiences of various types of assault and victimization (Tyler, Hoyt, Whitbeck, & Cauce, 2001). As a history of physical or sexual abuse is a strong correlate of criminal behavior (Baron, 2004, 2009), it is likely that this highly vulnerable group of maltreated young people would also engage in criminal behavior.
Reactions/Responses to Strains
Trauma symptoms/PTSD. According to general strain theorists, maltreatment and victimization are strains that may result in negative affective states, such as anger, depression, and anxiety (Baron, 2004; Baron & Hartnagel, 1997). The strains associated with past or current victimization contribute to these psychological challenges as evidenced by the elevated rates of mental disorders, such as PTSD, found among homeless youth (Thompson et al., 2006, 2007). Previous research has suggested that the psychological health challenges of homeless youth are linked to criminal behaviors (Baron, 2004, 2009). As a highly service-disengaged population who frequently do not seek mental health treatment (Kipke et al., 1997), homeless youth can be especially prone to illegal acts, particularly when negative symptoms remain untreated (Silver, 2000).
Substance dependence. Dependence on and abuse of substances are clearly associated with criminal activity (Baron & Hartnagel, 1997; Gaetz & O’Grady, 2002). Young people who are addicted to drugs and embedded in a street lifestyle often turn to theft, property crimes, and drug trafficking to finance their addictions (Farabee, Shen, Hser, Grella, & Anglin, 2001). There is evidence that substance dependency increases with the length of time youth are homeless or estranged from traditional society (Johnson, Whitbeck, & Hoyt, 2005; Whitbeck, 2009). Associating with substance-abusing peers and disaffiliating from conventional institutions and pro-social supports may place homeless youth at a heightened risk for crime. Youth may also abuse substances to cope with the daily strains of homelessness; self-medicating and drinking or abusing substances to numb negative emotions are common (Baron, 2004). Reduced inhibitions as a result of substance abuse, combined with the need to finance their abuse, may increase the risk these young people will engage in criminal behaviors (McMorris, Tyler, Whitbeck, & Hoyt, 2002).
Survival strategies. Survival strategies, which are common among homeless youth to obtain resources while on the streets, serve as another response to the strains of living on the streets. As youth become embedded in a street lifestyle, they are often marginalized and excluded from the formal economy due to lack of housing, difficulty attending to personal hygiene, food insecurity, and societal stigma (Dachner & Tarasuk, 2002; Ferguson et al., 2011; Gaetz & O’Grady, 2002). With little means to gain formal employment and income, many respond by turning to marginally legal and illegal activities to generate income (Gaetz & O’Grady, 2002). Survival strategies include survival sex (i.e., participating in sexual acts in exchange for money, food, lodging, clothing, or drugs), pimping, pornography, panhandling, theft, selling blood or plasma, or conning others (Gaetz & O’Grady, 2002; Kipke et al., 1997). Young people who are involved in a street lifestyle with like-minded peers may use survival strategies to support their addictions, meet their subsistence needs, or conform to peer pressure (Baron, 2009; Farabee et al., 2001). These activities may serve as a gateway to more serious forms of crime, as previous research suggests urban youth often follow a developmental trajectory involving less serious criminal behavior preceding more serious criminal involvement (Tolan, Gorman-Smith, & Loeber, 2000). The response to the various strains of street life by relying on survival strategies introduces homeless youth to criminal peer groups and increases the risk for more serious criminal involvement (Whitbeck, 2009).
It is evident that multiple strains and responses to the strains interact in the lives of homeless youth to increase their likelihood of engaging in criminal behaviors. Based on the assumptions that homelessness is a criminogenic experience (Baron & Hartnagel, 1997) and marked by strains and responses to strains (Baron, 2004, 2008), we hypothesized that a greater history of arrests will be reported by youth who: 1) have been homeless longer; 2) are more transient; 3) have been victimized; 4) meet the criteria for PTSD; 5) are drug dependent; and 6) use survival strategies to earn an income. Further, based on the assumptions that these strains are interrelated (Whitbeck et al., 1999) and that social estrangement and life stressors can lead to illegal behaviors (McMorris et al., 2002; Silver, 2000), we speculated that select strains may indirectly predict arrest history, as mediated through additional strains and responses to strains. Specifically, transience and length of time homeless will indirectly predict arrest history, as mediated through victimization, PTSD, drug dependence, and survival strategies.
For this cross-sectional, comparative study of homeless youth, researchers from Los Angeles, California; Denver, Colorado; New Orleans, Louisiana; and St. Louis, Missouri secured participation from host agencies providing care to homeless young people. Our selection of agencies was based on our existing relationships with service providers and their commitment to host the study. The participating agencies in each city consisted of multi-service, non-profit organizations that offer homeless, runaway and at-risk young people a comprehensive system of care, including street outreach, short- and long-term shelters, health care, mental health counseling, spiritual ministry, educational and employment services, and basic subsistence items.
Sampling and Recruitment Procedures
We recruited participants during 2005 in St. Louis, during 2008 in Los Angeles and Denver, and from 2008-2009 in New Orleans. We added Los Angeles, Denver, and New Orleans as study sites several years after data collection in St. Louis in order to expand the study’s scope to include small, mid-size, and large cities with homeless youth.
Using convenience sampling, we recruited 188 homeless youths aged 18–24 from Los Angeles (n = 50), Denver (n = 50), New Orleans (n = 50) and St. Louis (n = 38) from shelters, drop-in centers, and street outreach using similar methods. We used nearly identical recruitment procedures across cities with minor variations due to services emphasized in each location (e.g., more crisis-shelter users in Los Angeles and New Orleans, more drop-in service users in Denver, and more outreach-service users in St. Louis). We considered youths to be homeless if they had spent at least two weeks away from home during the past month (Whitbeck, 2009). To participate, youths had to meet three inclusion criteria: 1) be 18–24 years old, 2) have spent at least two weeks away from home in the month before the interview, and 3) provide written informed consent. We excluded young people if they were incapable of comprehending the consent form. We used a screening form to verify the participants’ ages and length of time away from home (i.e., that they had been away from home for at least two weeks).
Data Collection and Measures
Researchers and trained research assistants administered a 45- to 90-minute semi-structured retrospective interview to examine runaway history, transience, survival strategies, substance abuse, victimization, trauma symptoms, and arrest history among homeless youths. The researchers and research assistants conducted all interviews in private rooms at each host agency. We compensated the youths either $10.00 or an equivalent in gifts for their participation in the interview. Each investigator received human subjects’ approval from his or her respective university.
Dependent variable. We assessed arrest history by asking youths whether they had ever been arrested for nine types of criminal behaviors, including status offenses (curfew, under-age drinking, disorderly conduct, and so on), alcohol-related offenses, possession of illegal drugs, sale of drugs, violence (robbery, mugging, or rape), fighting or threatening with a weapon, theft (stealing property that did not belong to them), deception or forgery (writing “hot” checks), and vandalism (destruction of property) (coded 0 = no or 1= yes). Because different types of criminal behavior elicit more serious consequences in the criminal justice system than others and may indicate more severe delinquency, we created a severity index modeled on previous work measuring the severity of adolescent substance-use behavior (Wall & Kohl, 2007). We used this index previously to measure delinquency among at-risk young people (Bender, in press) and arrest activity among homeless youths (Ferguson, Bender, Thompson, Xie, & Pollio, in press). We created the severity index by assigning a value to each type of criminal behavior according to the severity of the crime and the likely consequence in the criminal justice system. We assigned a ‘0’ for no arrest; ‘1’ for a minor offense (status offenses); ‘2’ for each moderate offense [(1) alcohol-related offenses, (2) theft, (3) deception/forgery, (4) vandalism]; and ‘3’ for each serious offense [(1) possession or use of illegal drugs, (2) sale of drugs, (3) fighting or threatening with a weapon, (4) violent crime]. We then calculated arrest history as the summed score for all nine items indexed according to severity. Larger values denote greater arrest severity (range: 0–21; [1 × 1] + [2 × 4] + [3 × 4]).
Predictors of arrests. We asked the youths demographic information, such as their age, gender (0 = female, 1 = male), and race (0 = other, 1 = White). We determined youths’ length of time homeless from the number of months since they had left home for the first time for at least one night without parental supervision. We measured transience as the number of times the youths had moved between cities since leaving home for the first time.
We measured victimization by asking the youths whether in the previous six months they had ever been physically assaulted (other than sexual assault), sexually assaulted, or robbed (0 = no, 1 = yes). We formed a composite-score variable from the sum of these three items indicating the number of types of victimization the youths had ever experienced.
We measured survival strategies by inquiring whether the youths had received income during the prior six months from five non-traditional sources: panhandling, theft, prostitution, selling drugs, or selling blood/plasma. As a predictor of criminal arrests, this variable gauged whether the youths had received income from legal and illegal survival strategies, not whether they had been arrested for such activities. We measured all responses as 0 = no or 1 = yes. We created a composite measure by summing these five items.
We asked participants about their drug dependence using the Mini International Neuro-psychiatric Interview (MINI) (Sheehan et al., 1998). The MINI asks a series of dichotomous (no/yes) screening and symptom questions for drug dependence. A sufficient number of both positive responses to symptom questions and affirmative answers to screening questions is required to meet criteria for a diagnosis of substance dependence (Sheehan et al., 1998). We coded drug dependence as 0 = does not meet criteria for dependence or 1 = meets criteria. We also measured PTSD using the MINI. Similar to drug dependence, PTSD was a dichotomous variable (0 = no, 1 = yes) that measured whether the youths met the symptom criteria for this diagnosis.
We conducted descriptive analyses to depict the youths’ demographic characteristics. We used chi-square and ANOVA tests to examine city-level differences on the study predictors and outcome variables. Using Mplus, we tested the proposed theoretical model (Figure 1) via observed-variable path analysis using maximum likelihood parameter estimation. All proposed paths were based on general strain theory and the extant evidence outlined in our literature review suggesting the associations between and among variables. We performed transformation of the metric for the variable “length of time homeless” by dividing the original value by 10 to compensate for the larger metric relative to the variables “drug dependence” and “arrest history,” which we measured in smaller metrics. We determined model fit using conventional thresholds for the comparative fit index (CFI [> 0.90]) and root mean square error of approximation (RMSEA [< 0.05]) with 90% confidence interval (Muthen & Muthen, 2001).
We used Mplus to determine the significance of the overall mediation effects between transience/length of time homeless and victimization/PTSD/drug dependence/survival strategies and arrest history (Muthen & Muthen, 2001). We conducted the bootstrapping re-sample technique to handle the possible non-normal distribution of indirect effects in the mediation analysis. This technique is commonly used when sample sizes are small (Shrout & Bolger, 2002). We calculated the proportion of the mediation effect out of the total effect as the mediation effect divided by the total effect multiplied by 100%. We estimated the total effect of the predictors on the outcome variable in a second Mplus model with all mediators removed.
Figure 1. Proposed path model of arrest history among homeless youths, in relation to general strain predictors.
Full sample. Table 1 presents the demographic characteristics of the sample (N = 188). The mean age of participants was 20.26 (SD = 1.72) years. The majority (61.7%) were male. Roughly 45% were Black, 22% White, and 11% Latino.
Table 1. Demographic Characteristics for Full Sample
|Full Sample (N=188)|
|Length of Time Homeless (months)||61.77||47.90|
|Transience (# of moves between cities)||3.30||3.88|
|Victimization (total of 3 types)||0.50||0.75|
|Meets Criteria for PTSD
|Meets Criteria for Drug Dependence||43||22.9|
|Survival Strategies (panhandle, theft, prostitution, sell drugs, sell blood)||0.79||1.12|
|Arrest History (mean severity score)||3.13||3.48|
|Fighting/threatening with weapon||41||21.8|
|Violence (robbery, mugging, rape)||40||21.3|
|Sale of drugs||22||11.7|
|Possession of illegal drugs||7||3.7|
Youths averaged slightly more than five years (61.77 months) away from home. Since our sample included a heterogeneous group of homeless youth (e.g., runaway-homeless youth, throwaway youth, independent youth, emancipated foster youth, and so on) and since youth homelessness is rarely a one-time occurrence, categorical data (complementary to Table 1) on the youths’ length of homelessness (i.e., number of months since the youths left home for the first time without parental supervision) will help elucidate the general distribution of this variable. Close to one-fifth of our sample (17.8%) had been homeless for roughly less than 1 year (12.6 months); 16.6% had been homeless from 1 to 3 years (12.90–35.50 months); 16.6% from 3 to 5 years (37.10–60.30 months); 14.3% from 5 to 7 years (60.83–84.83 months); 13.9% from 7 to 9 years (86.2–108.13 months); and 13.8% for 9 or more years (109.50–249.87 months). The latter groups likely comprised youths who had grown up in institutional care and became homeless upon emancipating from the foster-care system, or homeless youths who often had multiple and repeated homeless experiences.
With respect to transience among our full sample, youths had made 3.30 inter-city moves since leaving home. Regarding arrest activity, 22.9% reported some kind of mild arrest activity; between 3.2% and 26.6% reported some kind of moderate arrest activity, and between 3.7% and 21.8% reported some kind of severe arrest activity.
Sub-sample by city. Table 2 includes descriptive statistics separately for each city sub-sample. We noted several differences among homeless youths across cities. For instance, regarding race/ethnicity, Los Angeles youths were more likely than youths in other cities to be Latino (24%, χ2 = 20.40, p = 0.000), whereas youths in Denver were predominantly White (42%, χ2 = 15.17, p = 0.000). New Orleans youths were more likely than their counterparts in other cities to be Black (68%, χ2 = 12.93, p = 0.000). With respect to transience, Los Angeles youths experienced a greater number of moves between cities since leaving home (mean = 5.18, SD = 4.75) than youths from other cities (F [1, 186] = 17.44, p = 0.000). Among the four cities, Denver youths were older (mean = 21.00, SD = 1.91) than youths from other cities (F [1, 186] = 13.62, p = 0.000). In addition, compared with youths in the other three cities, Denver young people experienced a more extensive arrest history overall (mean = 4.42, SD = 3.74, F [1, 183] = 9.89, p = 0.002). These young people also reported greater instances of physical assaults (40%, χ2 = 5.60, p = 0.018) as well as arrests for status offenses (36%, χ2 = 6.25, p = 0.012), theft (48%, χ2 = 15.28, p = 0.000), vandalism (22%, χ2 = 6.68, p = 0.010), and deception/forgery (8%, χ2 = 4.94, p = 0.026) than youths from the other three cities.
Table 2. Demographic Characteristics of Homeless Youths by City
n = 50
n = 50
n = 50
n = 38
|Fighting/threatening w/ weapon||10||(20.0)||13||(26.0)||10||(20.0)||8||(21.1)|
|Violence (robbery, mugging, rape)||13||(26.0)||15||(30.0)||6||(12.0)||6||(15.8)|
|Sale of drugs||5||(10.0)||7||(14.0)||6||(12.0)||4||(10.5)|
|Possession of illegal drugs||0||(0.0)||0||(0.0)||7||(14.0)||0||(0.0)|
|Length of Homelessnessa||48.64||(41.91)||63.52||(41.97)||76.87||(57.53)||61.21||(48.87)|
We conducted a path analysis on the full sample of 188 youths (a correlation matrix is available from Dr. Ferguson). Although prior studies using demographic sub-analyses suggest that homeless youths’ gender and race may influence outcomes (Whitbeck et al., 1999), we were limited by our sample size. Results of the hypothesized model revealed that nine path coefficients were significant (see Figure 2). Significant coefficients were found on the following paths: 1) from transience to victimization, 2) from victimization to PTSD, 3) from victimization to survival strategies, 4) from PTSD to drug dependence, 5) from drug dependence to survival strategies, 6) from drug dependence to arrest history, 7) from length of time homeless to drug dependence, 8) from length of time homeless to arrest history, and 9) from survival strategies to arrest history. The model represented an excellent fit to the data (χ2 = 11.01, df = 11, p = 0.44, CFI = 1.00, RMSEA = 0.002 [90% confidence interval = 0.000–0.073], PCFI = 0.393). Standardized path coefficients ranged from 0.14 to 0.37 and were significant at the p < 0.05 level. Specifically, length of time homeless and PTSD explained 8% of the variance in drug dependence. Victimization accounted for 14% of the variance in PTSD. Being victimized and drug dependent accounted for 13% of the variance in survival strategies. Collectively, length of time homeless, drug dependence, and survival strategies explained 17% of the variance in arrest history.
Figure 2. Standardized (unstandardized) parameter estimates for final path model of arrest history among homeless youths.
In response to our hypotheses, three of the six relationships had a direct effect on arrest history. Length of time homeless (β = 0.137, p = 0.050), drug dependence (β = 0.277, p = 0.001), and use of survival strategies (β = 0.171, p = 0.016) each significantly predicted arrest history.
Analyses of the overall mediation effects from transience to arrest history and from length of time homeless to arrest history reveal two mediation effects, one of which was significant at the p < 0.05 level and one at the p < 0.10 level. First, the overall mediation effect from transience to arrest history was 0.010 (p = 0.044). That is, 18.2% of the total variance in arrest history was explained by transience, victimization, PTSD, drug dependence, and survival strategies. In this case, youths who were highly transient were more likely than those who were less transient to have experienced more types of victimization and to meet criteria for PTSD, which contributed to being substance dependent and engaged in a greater number of survival strategies and, ultimately, to have a more extensive arrest history. Second, the overall mediation effect from length of time homeless to arrest history was 0.036 (p = 0.055). That is, 27.1% of the total variance in arrest history was accounted for by length of time homeless, drug dependence, and survival strategies. Youths who had been on the streets longer were more likely to be dependent on substances, which contributed to a greater number of survival strategies and ultimately a more extensive arrest history.
Our final model, illustrated in Figure 2, represents an improvement over our proposed theoretical model depicted in Figure 1. To arrive at this final model, we considered the likelihood of alternative models to explain arrest history by testing the original theoretical model, containing four additional paths between variables based on general strain theory, and the empirical precedents outlined in our literature review. To test our six hypotheses, we compared these competing models containing paths from length of time homeless, transience, victimization, PTSD, drug dependence, and survival strategies to arrest history. We originally deemed these paths important, since extant evidence suggests that criminal activity may be more likely among homeless youths who have been homeless longer (Baron & Hartnagel, 1997), who are highly mobile and lack connections with trusted peers and adults (Bender et al., 2007), who have experienced a history of abusive relationships (Baron, 2004, 2009), who have psychological health challenges (Silver, 2000), who are drug dependent (Farabee et al., 2001), and who use survival strategies to meet their needs (Gaetz & O’Grady, 2002). The three models with insignificant paths (transience to arrest history, victimization to arrest history, and PTSD to arrest history) did not improve overall model fit using the likelihood ratio test of ΔX
This study aimed to explore the interactions among mental health and situational strains—and responses to such strains—that are associated with arrest history among homeless youth. Two major types of strain (Agnew, 1992) are supported by data from this study. First, strain may result from the presence of negative stimuli. In this case, delinquency may be a means of alleviating the strain by escaping from or coping with the negative stimuli or seeking revenge against its source. Likewise, strain may result from the failure to achieve positively valued goals. Delinquent behavior in this case may be a method for illegally attaining one’s goals or coping with one’s failed expectations.
With respect to the first type of strain, homeless youths commonly encounter multiple, and often chronic, negative stimuli in their daily lives on the streets. Living on the streets often requires youths to sleep alone at night in dangerous areas or to move frequently to find safer locations (Whitbeck, 2009). Homeless youths also experience frequent victimization and trauma once on the streets (Tyler et al., 2001). Many youths on the streets engage in risky behaviors to cope with the daily stressors of street living (e.g., drug use) as well as to generate income for survival (e.g., stealing, drug dealing, prostitution) (Kipke et al., 1997). Drawing from principles of general strain theory, homeless youths who spend extended periods of time on the streets surrounded by multiple and often chronic strains may engage in criminal behavior in an effort to avoid or alleviate the strains (e.g., committing alcohol-related offenses, drug possession/use, theft) or to seek revenge against the source of the strains (e.g., committing violent offenses, fighting with a weapon, vandalism) (Agnew, 1992). Our findings support this explanation: Youths who had been homeless longer were more likely to be both dependent on drugs and to have more extensive arrest histories.
Our mediation analysis findings further demonstrate a complex process in which youths who were more transient were more likely to experience victimization, meet the criteria for PTSD and drug dependence, and to use survival strategies, all of which place them at increased risk for criminality. In this case, highly mobile youths may be presented with increased negative stimuli as they move between cities, which can ultimately result in criminal activity. Highly transient youths may be more susceptible to experiencing victimization on the streets, since repeated moves between cities limit homeless youths’ awareness of local safe havens (Ferguson et al., 2011). As a result of victimization, homeless youths may develop symptoms of mental illness, such as PTSD (Thompson et al., 2006). Substance dependence may, in turn, occur as a coping mechanism to numb or escape from emotions associated with trauma (Baron, 2004). These accumulating strains among homeless youths are associated with greater involvement in a street lifestyle and use of survival strategies to support addictions or to meet subsistence needs (Baron, 2009). Extant evidence suggests that substance abuse, survival strategies, and status offenses serve as gateways to more serious forms of crime (Baron & Hartnagel, 1997; Humphrey, 2004; Tolan et al., 2000). Homeless youths may thus turn initially to less severe forms of survival strategies but later progress to more serious criminal activity to satisfy their needs.
Furthermore, our findings support a second major type of strain in the lives of homeless youths—the disjunction between their expectations and their actual achievements. In this case, homeless youths may experience an increasing disparity between their expectations and their actual achievements the longer they spend on the streets. This discrepancy between their desired and actual outcomes may lead to negative coping responses and, ultimately, engagement in criminal activity. Extended homelessness can be considered a strain that makes traditional expectations (e.g., academic achievement, economic self-sufficiency, and so on) more difficult to attain. Prior research suggests that increased length of time on the streets is associated with higher rates of academic drop-out and unemployment (Baron, 1999; Baron & Hartnagel, 1997) as well as greater drug use and involvement with deviant peer groups (Fest, 2003). Due to the presence of these strains, homeless youths may use illicit behaviors as a means of attaining goals such as independent living and economic self sufficiency, or as a means of coping with failed expectations (Gaetz & O’Grady, 2002).
In this study, we observed an overall mediation effect (albeit significant at the p = 0.10 level) in support of this explanation. The longer youths were homeless, the more likely they were to be dependent on drugs, to use survival strategies to generate income and, ultimately, to have a more extensive arrest history. Youths may have used substances to cope with failed expectations associated with their homeless status (e.g., unemployment, academic drop-out, precarious housing). Their heightened drug dependence then further contributed to a need to finance their addictions and to earn an income to meet their needs. Ultimately, homeless youths may have resorted to criminal activity as a means of expressing their anger, dissatisfaction, and disappointment with their failed expectations (e.g., in committing disorderly conduct, vandalism, or violent acts) or of achieving their expectations after all via illicit behaviors (e.g., in committing forgery, theft, and drug sales).
The study findings and conclusions drawn from them should be interpreted with caution because of several limitations. Perhaps most importantly, due to the cross-sectional nature of this study, the directions and order of the hypothesized relationships were based on theory, not on temporal order. It is thus important to acknowledge that even though our findings revealed significant associations between predictors and arrest history, causality cannot be drawn from these observed associations given the cross-sectional data used in this study.
Second, although the methodology allows comparisons among cities, the samples are not representative of the populations in any of the four cities. Due to the use of a convenience sample, we cannot exclude the possibility of volunteer bias. Because homeless youths are transient and difficult to locate, probability sampling is often not feasible. Instead, purposive sampling methods through street locations and service agencies are commonly implemented in empirical investigations with homeless young people (Clatts, Davis, & Atillasoy, 1995). Differences in the ability of this sampling method to saturate the available population within cities also remain unclear. Likewise, the choice of the four cities in this study was based on feasibility, rather than on representativeness of various types of homeless youth. Because this is among the first studies to compare homeless youth across the nation, it is impossible to determine whether other cities fit the pattern of these findings. Nevertheless, the data collected in this study from four disparate U.S. cities lend credibility and generalizability to the findings.
It should also be noted that the correlates included in this study were proxies representing different types of general strain in youths’ lives. Central measures of general strain theory—including affective measures such as anger, ability to achieve positively valued goals, and expectations for the future—would have been preferable to gain a more accurate understanding of the influence of general strain on homeless youths’ criminality. It is possible the influence of general strain was underestimated without these key indicators, and future research should investigate their impact on criminal behavior in this population. Similarly, other general strain proxy variables were also not collected in this study. For example, association with deviant peer groups on the streets has been related to greater endorsement of the street culture and greater criminality (Kipke et al., 1997). As this variable was not collected in this study, future research should examine peer influences in the context of criminal behavior among homeless youth.
Finally, the outcome measure of arrest history was limited in three important ways. First, we gathered arrest information through self-reports by the youths and did not corroborate this information with official records. The youths may have underreported arrest data because they were reticent to convey sensitive information about their illegal behaviors to adults. However, a notable strength was that the interviewers had considerable histories as researchers, staff, or volunteers with homeless-youth organizations and were familiar with the street lifestyle. Because the interviewers were known and trusted by many of the subjects, it was less likely that the youths would bias their responses. Second, in collecting data on the number of arrests for illegal behaviors (as opposed to offenses committed, regardless of arrests), we may have captured a more conservative estimate of criminal history, which may have minimized the relationships found.
Third, we noted city-level differences on the outcome variable, as well as on two correlates (transience and physical assault), both of which may have biased our results. In comparison with youths from the other cities, Denver youths were significantly older and experienced a more extensive arrest history. In this case, the youths’ older age may explain their more extensive arrest history, as previous research suggests that less severe criminal behavior (e.g., status offenses) in younger youths may serve as a gateway to more serious forms of crime as youths age (Tolan et al., 2000). Heterogeneity among the four-city sample in demographics and other study variables can be expected, as prior research with homeless youths using multi-city datasets demonstrates that youths in different regions of the United States vary significantly in their runaway behaviors, transience, substance use, suicidal ideation, and reports of physical abuse and sexual abuse (Ferguson, 2010; Thompson, Maguin, & Pollio, 2003). It is also likely that city-level differences existed in both contexts and patterns of crime, yet we did not collect more extensive crime data in this study. Future studies with this population would benefit from further scrutiny of how homeless youths’ patterns, types, and frequencies of crime are influenced by different geographical contexts (e.g., small vs. large, urban vs. suburban vs. rural, more-affluent vs. less affluent).
Implications for Research and Practice
Despite these limitations, this study offers one of the first applications of general strain theory to identify both mental health and situational strains—and responses to strains—among homeless youth. Findings have important implications for research and preventive interventions to address criminal activity among this population. First, findings supporting general strain theory as a useful explanatory framework for homeless youths’ arrest behavior suggest that further research is warranted to develop effective interventions to protect youth while homeless by minimizing the daily stressors associated with a street lifestyle. It is important for homeless youth researchers to partner with service providers in identifying the trajectories of homeless youth and their criminal involvement. Greater understanding of the risk profiles of homeless youth who are more likely to offend—as well as protective factors that inhibit offending behaviors—will inform prevention and intervention efforts aimed at reducing further criminal activity among this population. Future studies should examine whether criminality can be reduced through protective strategies that establish stable housing, limit youths’ geographic transience, increase access to formal employment, keep youth safe while homeless, and enhance substance abuse and mental health treatment.
Second, efforts to stabilize and house youth in one community, in which they can reduce the strains associated with transience and length of homelessness, are likely to be associated with reduced arrest rates. While criminal behavior and arrests appear common among these youth, development of services to connect with and protect this population is likely to help them stay safe while homeless and avoid the emotional and behavioral reactions to strains that are associated with criminal behavior. This may be especially challenging as homeless youth frequently are estranged from formal employment, housing, and educational systems, as well as disconnected from their families and adult role models. To better engage with and protect these youth, providing housing and other residential accommodations to geographically stabilize this population will enable them to establish fixed relationships with supportive adults and institutions and to pursue mental health and substance abuse treatment. Alternatively, use of virtual/online harm-reduction interventions (e.g., using social networking software) for highly mobile homeless youth may provide portable safety strategies amid their transient episodes.
Furthermore, this study identifies several mental health correlates of arrests by homeless youth. Greater attention to the mental health status of homeless youth within the juvenile justice system could likely help prevent future offending behaviors (Kosterman, Graham, Hawkins, Catalano, & Herrenkohl, 2001). Findings by Kempf-Leonard and Johansson (2007) reveal that youths arrested for running away from home (particularly female runaways) are often not offered any form of intervention to help them cope with their elevated levels of mental illness and histories of abuse and victimization. Instead, these youths frequently receive a brief warning and are sent home. Earlier intervention responses within the juvenile justice system may aid these youths in reducing correlates identified in this study (e.g., PTSD and drug abuse and dependence) as risk factors for continued criminal activity and arrests. For instance, evidence-based interventions that have demonstrated effectiveness in reducing substance abuse and other risky behaviors include cognitive-behavioral therapy, motivational interviewing, and health and risk-reduction information and education (Baer et al., 2004). These and other evidence-based interventions are likely to reduce risk to the youths and to society as a whole.
About the Authors
Kristin M. Ferguson, PhD, is Associate Professor at the Silberman School of Social Work at Hunter College, City University of New York. Her research interests include homeless and street-living youth, social capital, criminal behavior among street youth, and the effectiveness of vocational interventions for homeless youth.
Kimberly Bender, PhD, is Assistant Professor at the University of Denver School of Social Work, where she studies the etiology of juvenile delinquency, interventions for preventing criminal activity, and methods for improving juvenile well-being and psychosocial functioning.
Sanna J. Thompson, PhD, is Associate Professor at the University of Texas at Austin, School of Social Work. Her research focuses on high-risk youth and families, with special emphasis on runaway/homeless youth.
Bin Xie, PhD, is Associate Professor at the School of Community and Global Health, Claremont Graduate University. His research focuses on cultural, behavioral, and psychological aspects of smoking and obesity prevention in American and Chinese adolescents.
David Pollio, PhD, is Professor, Hill Crest Foundation Endowed Chair in Mental Health Research, University of Alabama School of Social Work. His research focuses on mental health and addiction services, homelessness, runaway and street youth, evidence-based practice, group interventions, and geographic information systems.
Agnew, R. (1992). Foundation for a general strain theory of crime and delinquency. Criminology, 30, 47–87.
Baer, J. S., Peterson, P. L., & Wells, E. A. (2004). Rationale and design of a brief substance use intervention for homeless adolescents. Addiction Research & Theory, 12(4), 317–334.
Baron, S. W. (1999). Street youths and substance use: The role of background, street lifestyle, and economic factors. Youth & Society, 31(1), 3–26.
Baron, S. W. (2004). General strain, street youth and crime: A test of Agnew’s revised theory. Criminology, 42, 457–483.
Baron, S.W. (2008). Street youth, unemployment, and crime: Is it that simple? Using general strain theory to untangle the relationship. Canadian Journal of Criminology and Criminal Justice, 50(4), 399–434.
Baron, S. W. (2009). Differential coercion, street youth, and violent crime. Criminology, 47, 239–268.
Baron, S. W., & Hartnagel, T. F. (1997). Attributions, affect, and crime: Street youths’ reactions to unemployment. Criminology, 35(3), 409–434.
Bender, K. (In press). The mediating effect of school engagement in the relationship between child maltreatment and delinquency. Children and Schools.
Bender, K., Thompson, S. J., McManus, H., Lantry, J., & Flynn, P. M. (2007). Capacity for survival: Exploring strengths of homeless street youth. Child and Youth Care Forum, 36, 25–42.
Clatts, M. C., Davis, W. R., & Atillasoy, A. (1995). Hitting a moving target: The use of ethnographic methods in the evaluation of AIDS outreach programs for homeless youth in NYC. Qualitative Methods in Drug Abuse and HIV Research: National Institute on Drug Abuse Research Monograph, 157, 117–135.
Dachner, N., & Tarasuk, V. (2002). Homeless ‘squeegee kids’: Food insecurity and daily survival. Social Science and Medicine, 54, 1039–1049.
Farabee, D., Shen, H., Hser, Y-I., Grella, C. E., & Anglin, M. D. (2001). The effect of drug treatment on criminal behavior among adolescents in DATOS-A. Journal of Adolescent Research, 16(6), 679–690.
Ferguson, K., Bender, K., Thompson, S., Xie, B., & Pollio, D. (2011). Correlates of street-survival behaviors in homeless young adults in four U.S. cities. American Journal of Orthopsychiatry, 81(3), 401–409. DOI 10.1111/j.1939-0025.2011.01108.x
Ferguson, K., Bender, K., Thompson, S., Xie, B., & Pollio, D. (In press). Exploration of arrest activity among homeless young adults in four U.S. cities. Social Work Research.
Ferguson, K., Jun, J., Bender, K., Thompson, S., & Pollio, D. (2010). A comparison of addiction and transience among street youth: Los Angeles, California, Austin, Texas, and St. Louis, Missouri. Community Mental Health Journal, 46(3), 296–307.
Fest, J. (2003). Understanding street culture: A prevention perspective. School Nurse News, 20, 16–18.
Gaetz, S., & O’Grady, B. (2002). Making money: Exploring the economy of young homeless workers. Work, Employment & Society, 16(3), 33–456.
Halcon, L. L., & Lifson, A. R. (2004). Prevalence and predictors of sexual risks among homeless youth. Journal of Youth and Adolescence, 33, 71–80.
Humphrey, A. (2004). Criminalization of survival attempts: Locking up female runaways and other status offenders. Hastings Women’s Law Journal, 15, 165–184.
Johnson, K. D., Whitbeck, L. B., & Hoyt, D. R. (2005). Substance abuse disorders among homeless and runaway adolescents. Journal of Drug Issues, 35(4), 799–816.
Kempf-Leonard, K., & Johansson, P. (2007). Gender and runaways: Risk factors, delinquency, and juvenile justice experiences. Youth Violence and Juvenile Justice, 5, 308–327.
Kipke, M. D., Unger, J. B., O’Connor, S., Palmer, R. F., & LaFrance, S. (1997). Street youth, their peer group affiliation and differences according to residential status, subsistence patterns, and use of services. Adolescence, 32(127), 655–669.
Kosterman, R., Graham, J. W., Hawkins, J. D., Catalano, R. F., & Herrenkohl, T. I. (2001). Childhood risk factors for persistence of violence in the transition to adulthood: A social development perspective. Violence and Victims, 16(4), 355–369.
Kushel, M. B., Yen, I. H., Gee, L., & Courtney, M. E. (2007). Homelessness and health care access after emancipation: Results from the Midwest evaluation of adult functioning of former foster youth. Archives of Pediatric and Adolescent Medicine, 161(10), 986–993.
McMorris, B. J., Tyler, K. A., Whitbeck, L. B., & Hoyt, D. R. (2002). Familial and “on-the-street” risk factors associated with alcohol use among homeless and runaway adolescents. Journal of Studies on Alcohol, 63(1), 34–43.
Muthen, L., & Muthen, B. O. (2001). Mplus user’s guide (second edition). Los Angeles, CA: Muthen & Muthen.
O’Grady, B., & Gaetz, S. (2004). Homelessness, gender and subsistence: The case of Toronto street youth. Journal of Youth Studies, 7, 397–416.
Sheehan, D. V., Lecrubier, Y., Sheehan, K. H., Amorim, P., Janavs, J., Weiller, E., Hergueta, T., Baker, R., & Dunbar, G. C. (1998). The Mini-International Neuropsychiatric Interview (M.I.N.I.): The development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. Journal of Clinical Psychiatry, 59(Suppl. 20), 22–33.
Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and non-experimental studies: New procedures and recommendations. Psychological Methods, 7, 422–445.
Silver, E. (2000). Extending social disorganization theory: A multilevel approach to the study of violence among persons with mental illness. Criminology, 38(4), 1043–1074.
Thompson, S., Jun, J., Bender, K., Ferguson, K. M., & Pollio, D. E. (2010). Estrangement factors associated with addiction to alcohol and drugs among youth in three U.S. cities. Evaluation and Program Planning, 33, 418–427.
Thompson, S. J., Maccio, E., Desselle, S., Zittel-Palamara, K. (2007). Predictors of traumatic distress among runway youth utilizing two service sectors. Journal of Traumatic Stress, 20, 553–563.
Thompson, S. J., Maguin, E., & Pollio, D. E. (2003). National and regional differences among runaway youth using federally funded crisis shelters. Journal of Social Service Research, 30(1), 1–17.
Thompson, S. J., McManus, H. H., & Voss, T. (2006). PTSD and substance abuse among youth who are homeless: Treatment issues and implications. Brief Treatment and Crisis Intervention, 6(3), 206–217.
Tolan, P. H., Gorman-Smith, D., & Loeber, R. (2000). Developmental timing of onsets of disruptive behaviors and later delinquency of inner-city youth. Journal of Child and Family Studies, 9(2), 203–220.
Tyler, K. A., Hoyt, D. R., Whitbeck L. B., & Cauce, A. M. (2001). The effects of a high-risk environment on the sexual victimization of homeless and runaway youth. Violence and Victims, 16(4), 441–455.
Tyler, K. A., & Johnson, K. A. (2006). Pathways in and out of substance use among homeless emerging adults. Journal of Adolescent Research, 21(2), 133–157.
Wall, A. E., & Kohl, P. L. (2007). Substance use in maltreated youth: Findings from the National Survey of Child and Adolescent Well-Being. Child Maltreatment, 12(1), 20–30.
Whitbeck, L. B. (2009). Mental health and emerging adulthood among homeless young people. New York: Psychology Press.
Whitbeck, L. B., Hoyt, D. R., & Ackley, K. A. (1997). Abusive family backgrounds and later victimization among runaway and homeless adolescents. Journal of Research on Adolescence, 7(4), 375–392.
Whitbeck, L. B., Hoyt, D. R., & Yoder, K. A. (1999). A risk-amplification model of victimization and depressive symptoms among runaway and homeless adolescents. American Journal of Community Psychology, 27(2), 273–296.