Abstract
While there is a growing body of research on intimate partner violence (IPV) experienced by the housed youth population, a limited amount is known about IPV experienced by homeless youth. To our knowledge, no previous studies have examined how homeless youths’ experience of IPV is related to their social network, even though the social networks of homeless youth have been shown to be significant indicators of health and mental health. The purpose of this study is to understand the relationship between IPV, gender, and social networks among a sample of 386 homeless youth in Los Angeles, California. Results revealed that one fifth of the sample experienced IPV in the past year. Stratified regression models revealed that IPV was not significantly related to any measure of male social networks; however, females who experienced IPV had more male friends (β = 2.03, SE = 0.89, p < .05) than females who did not experience IPV. Female homeless youth who witnessed family violence during childhood had more male friends (β = 2.75, SE = 1.08, p < .05), but those who experienced sexual abuse during childhood had fewer male friends (β = −2.04, SE = 0.93, p < .05). Although there was no significant difference in the rate of IPV victimization across genders, the context of this abuse appears to be drastically different. The results suggest that females with more male friendships are at greater risk for exposure to IPV. To date, there are few effective youth-targeted IPV prevention programs and none have been shown to be effective with homeless youth. These results provide insight into future program development.
Keywords
Introduction
There are approximately 1.6 million homeless youth in the United States each year (Ringwalt, Greene, & Robertson, 1998; Toro, Dworsky, & Fowler, 2007). Youth who are homeless suffer from a vast number of health risks as a result of risk behaviors, limited access to resources and increased exposure to endangerment. Intimate partner violence (IPV) among homeless youth is a frequently experienced health risk (Slesnick, Erdem, Collins, Patton, & Buettner, 2010; Tyler & Beal, 2010), yet it has continuously been understudied within this population. Research with homeless youth over the past two decades have consistently implicated social networks as important determinants of their health, mental health, housing, and experiences of violence (Milburn et al., 2009; Rice, 2010; Rice, Stein, & Milburn, 2008; Unger et al., 1998; Wenzel et al., 2012). Work with housed populations has shown that experiences with IPV are often associated with adult social relationships (Coohey, 2007). The purpose of the current study is to understand how the experience of IPV is associated with the structure of an individual’s social network.
Background
The U.S. Centers for Disease Control and Prevention (CDC; 2010) defined youth IPV as physical, sexual, or psychological harm that is inflicted by a current or former intimate partner that occurs within the context of a youth dating relationship. IPV can impact youth regardless of race, culture, or socioeconomic status. However, certain groups, such as homeless youth, are at greater risk for experiencing IPV when compared with the general youth population (Slesnick et al., 2010; Tyler, Melander, & Noel, 2009). Homeless youth samples have reported that up to 73% of youth have experienced violence inflicted by an intimate partner in their lifetime (Slesnick et al., 2010; Tyler & Beal, 2010). It has been well documented that homeless youth experience high rates of physical and sexual victimization as a result of homelessness (Tyler, Whitbeck, Hoyt, & Cauce, 2004; Whitbeck & Hoyt, 1999), including violence inflicted by an intimate partner. A prior study found that 60% of homeless youth experienced physical violence in their most recent relationships (Boris, Heller, Sheperd, & Zeanah, 2002). They also found that on average youth had been in two violent relationships in their lifetime.
Risk Factors of IPV for Homeless Youth
An early history of childhood maltreatment puts this population at risk for IPV. Homeless youth experience high rates of neglect and physical and sexual abuse, and many witness domestic violence (Tyler & Johnson, 2006; Tyler et al., 2004; Whitbeck & Hoyt, 1999; S. M. Wolfe, Toro, & McCaskill, 1999). Approximately one third of homeless youth report experiencing childhood sexual abuse and up to 60% of homeless youth report experiencing physical abuse (Boris et al., 2002; Powers, Eckenrode, & Jaklitsch, 1990; Tyler & Cauce, 2002). Moreover, for many youth, escaping a negative family environment is a primary reason for becoming homeless (Milburn, Rotheram-Borus, Rice, Mallet, & Rosenthal, 2006; Whitbeck & Hoyt, 1999; Zide & Cherry, 1992).
Considering homeless youths’ overwhelming negative family histories, the prevalence of adolescent IPV is not surprising. In the general youth population, experience of childhood maltreatment is the most common predictor of later life victimization and perpetration of IPV (Ehrensaft et al., 2003; Fergusson, Boden, & Horwood, 2006, 2008; McCloskey & Lichter, 2003). Although the different forms of childhood maltreatment appear to have a cumulative effect (Hamby, Finkelhor, Turner, & Ormrod, 2010), the exposure to marital domestic violence may be the greatest independent risk for being a victim of any act of partner violence (Ehrensaft et al., 2003; Fergusson et al., 2006). Family adversity and childhood conduct disorders have also been found to significantly predict later life IPV victimization and perpetration (Fergusson, Boden, & Horwood, 2008). For homeless youth, childhood maltreatment is found to be related to IPV. Slesnick et al. (2010) also found that homeless youth who experienced physical or sexual victimization in childhood were twice as likely to be involved in violent intimate relationships.
Homeless youth are frequently exposed to risks and thus are more likely to experience violent victimization compared with the general adolescent population (Tyler et al., 2004; Whitbeck & Hoyt, 1999). Most youth spend a majority of their time in a street environment, thereby increasing their exposure to potential perpetrators. Violence is also a favored method of dispute resolution for homeless youth on the streets (Baron, 2003; Tyler & Beal, 2010). Peers’ involvement in violence, as well as risky subsistence strategies, criminal delinquency, and the use of illicit drugs and alcohol increase the risk of both physical and sexual victimization (Baron, 2003; Tyler et al., 2004).
Consequences of IPV for Homeless Youth
Previous literature has highlighted the importance of a homeless youth’s peer network and how it is directly related to a youth’s experience on the streets. Different types of relationships (i.e., street-based peers, home-based peers, friends, relatives, intimate partners, etc.) can be sources of either risk or resiliency. For homeless youth with connections to pro-social ties (e.g., peers with jobs or connections to family), it has been found that they show less anti-social behaviors, less substance use, an increase in HIV prevention behaviors including condom use, and an increased likelihood to successfully exit homelessness (Milburn et al., 2009; Rice, 2010; Rice et al., 2008). Connections to street-based peers embedded in dense networks with other risk-taking homeless youth increase the risk for anxiety and depression (Rice, Kurzban, & Ray, 2012), substance use (Rice, Milburn, & Rotheram-Borus, 2007; Rice, Milburn, Rotheram-Borus, Mallett, & Rosenthal, 2005; Unger et al., 1998), and increased sexual risk-taking (Martino et al., 2011; Rice, Monro, Adhikari, & Young, 2010). Maintaining connections to home-based ties appears to be a protective factor for improved mental health, reduced substance use, and reduced sexual risk-taking (Rice et al., 2012; Rice, Milburn, & Monro, 2011; Rice et al., 2010).
Research in housed adult populations has shown that an individual’s close relationships can be greatly impacted by IPV (Coohey, 2007; Katerndahl, Burge, Ferrer, Becho, & Wood, 2012; Levendosky et al., 2004). The social networks of victims of IPV were significantly smaller and offer less support than non-victims (Katerndahl et al., 2012) and abuse severity is negatively related to frequency of contact with friend and family ties (Coohey, 2007). As abuse becomes more severe, the frequency of contact decreases. This suggests that an individual’s social network may be a target for batterers using isolative tactics and limiting the victim’s access to social support mechanisms. It is also possible that this reflects how continued exposure to violence affects the victim’s ability to maintain quality relationships outside of the violent relationship. Previous literature in the general adolescent population has shown that various social network properties increase the risk for experiencing IPV. Adolescents in small, dense, primarily male peer networks with higher levels of delinquency are more likely to be perpetrators of IPV (Casey & Beadnell, 2010). In addition, low self-esteem and insufficient social skills are a risk for both experiencing and perpetrating IPV (Camacho, Ehrensaft, & Cohen, 2012).
Gender and Peer Relationships
Several pieces of evidence suggest that the high-risk environment of the street has a different effect of victimization for females compared with males. Female homeless youth are more likely to be victims of sexual assault even though they engage in similar high-risk behaviors as male homeless youth (Tyler et al., 2004). Female homeless youth are also twice as likely to be victims of IPV (Slesnick et al., 2010). Gaetz (2004) suggested that, in fact, the streets are a gendered environment and when confronting a lack of resources, female homeless youth may establish relationships that provide shelter or income that oftentimes increases the risk of assault and exploitation. Risky street lifestyle and marginalized social status of homeless youth also may result in less access to supportive resources including those that are aimed at preventing or intervening with partner violence (Gaetz, 2004). Without access to supportive resources, formal or informal, it is more likely that a youth will remain in a violent relationship rather than terminate it.
Outside of the context of youth homelessness, a consideration of differences across gender is necessary when determining the effects of IPV on peer relations, as there are critical differences in the structure of boys’ and girls’ peer interaction patterns. In a comprehensive review, Rose and Rudolph (2006) found that these differences are most salient during adolescence. In this period, girls and boys interact with same-sex peers more frequently than opposite sex peers. Boys are more likely to interact in a larger peer group that is dense with internal connections. Girls are more likely to maintain dyadic friendships and are more pro-social, reporting more self-disclosure and empathy in friendships than boys.
Current Study
To understand how IPV is related to a homeless youth’s peer relationships, this research will utilize a gender-stratified analysis and egocentric social network data. It is hypothesized that male and female homeless youth who experienced IPV in the past 12 months will have different social networks than those who have not experienced partner violence. It is hypothesized that the effect of IPV will be different for females compared with males as a result of gender differences in peer relational styles. This study will focus on the associations between IPV and the structure of an individual’s social network.
Method
A sample of 406 homeless youth (aged 13-25 years) currently accessing services from two homeless youth day-service drop-in centers, one in Hollywood, California, and one in Santa Monica, California, were recruited in two waves during the months of October 2011 and February 2012. The agencies selected provided weekday services to eligible homeless youth that include basic needs, medical, mental health, case management as well as referrals and connections to other programs such as housing. All youth accessing these agencies during the data collection period were invited to participate. In Santa Monica, 93.3% of the population was interviewed, and in Hollywood, 80.2% of the population was interviewed. In total, 361 youth completed both parts of the survey. A consistent set of two research staff members were responsible for all recruitment to prevent youth completing the survey multiple times within each data collection period per site.
Procedures
Signed voluntary informed consent was obtained from each youth. The Institutional Review Board at the University of Southern California approved all survey items and procedures as well as waived parental consent for minors who may not have a parent or adult. The study consisted of two parts: a computerized self-administered survey and a social network interview. The computerized survey included an audio-assisted version for those with low literacy and could be completed in English or Spanish. The computerized survey included approximately 200 questions and took an hour to complete on average. The social network interview was conducted by a trained research staff member as per the method used in Rice’s (2010) study. Interviewers first explained to the participant that they were collecting information about everyone in the youth’s social network in the previous 30 days. Youth were asked to name every person they have interacted with either face-to-face, on the phone, or in written communication, including text message, email, or through a social networking site like Facebook, Twitter, and so on. Interviewers used a large sheet of paper and colored markers to create a network map for each participant. Individuals that the youth nominated to include in their social network were referred to as “alters” and were written around the edge of the paper in an arc around individuals name that was written in the center of the paper. After the youth was finished nominating alters, the interviewer went through a series of approximately 60 questions regarding the different attributes of each alter (e.g., gender, race/ethnicity, age, years known, relation to participant). The interview took between 15 and 30 min to complete depending on the participants network size. All participants received US$20 in cash or gift cards as compensation for their time.
Measures
All responses were based on self-reports. Demographic items included gender, sexual orientation, race/ethnicity, age, and time homeless (participants were asked how long in total they have been homeless, in years and months).
IPV
Each participant was asked during the computerized self-administered survey, “During the past 12 months, did your boyfriend or girlfriend ever hit, slap, or physically hurt you on purpose?” This question was adopted from Youth Behavioral Risk Study (CDC, 2010). Participants’ answers were dichotomized into a dummy variable (no = “0”; yes = “1”).
Childhood trauma
Three variables were created to control for experience of childhood trauma. Physical abuse and witness of family violence were assessed by asking each participant if they had ever been hit, punched, or kicked very hard at home or witnessed a family member being hit, punched, or kicked very hard at home. Sexual abuse was assessed by asking the participant if he or she had experienced having an adult or someone much older touch your private sexual body parts when you did not want them to. All statements were adapted from the University of California at Los Angeles post-traumatic stress disorder reaction index for the Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; American Psychiatric Association, 1994; Pynoos, Steinberg, & Rodriguez, 1999).
Social network variables for peer relationships
For each nominated alter, the participant was asked who they considered to be a friend. All social network variables were prompted as reflections of the previous 30 days based on the previous methods used in the Rice (2010) study. A variable was created for the total number of friends in a network. Variables were also created for the total number of street-based friends and the total number of home-based friends. Street-based friends are alters the individual identified as a friend and one who has unstable housing, is currently homeless or lives in a shelter. Home-based friends are alters the individual identified as a friend and who he or she knew before he or she became homeless. In addition, variables were created to identify the number of male and female friends each youth identified in his or her network. After initial analyses, only the variable for male friends was pursued further as a dependent variable because it was found that female friends were not significantly related to any of the independent variables of interest.
Analyses
The analysis occurred in two stages. First, chi-square tests and independent sample t tests were used to determine any significant differences in childhood trauma variables and the dependent social network variables when comparing females and males. For the second stage of the analysis, the sample was stratified by gender. Six participants identified as transgender and were placed in the gender category that they currently identified with (i.e., male-to-female in the female category; female-to-male in the male category). Multivariate ordinary least square (OLS) regression equations were then examined with each social network variable as the dependent variable for each gender-based sample, making a total of eight unique models (i.e., for males and females, separate models were run for the following social network variables: total number of friends, number of street-based friends, number of home-based friends, and number of male friends). The final models also included demographic control variables present in the previous literature (Milburn et al., 2006; Slesnick et al., 2010; Wenzel, Tucker, Golinelli, Green, & Zhou, 2010), including age, race, time homeless, and sexual orientation as well as experience of childhood physical or sexual abuse, and witnessing family violence.
Results
Descriptive statistics for the two gender groups and the overall sample are presented in Table 1. A total of 386 youth participated in the self-survey portion of the interview. The sample was primarily male. Approximately, three quarters of the sample were male and one quarter was female. One quarter of the sample responded as being either homosexual, bisexual, queer, or questioning. Participants’ average age was 21.4 years old. The average time spent homeless was just below 3 years.
Descriptive Statistics Stratified by Gender.
More than one fifth of the sample experienced violence inflicted by an intimate partner in the past year. Chi-square results shown in Table 2 show no significant differences in rates between females experiencing. Overall, 44% of participants experienced physical abuse inflicted by a family member as a child and 39% witnessed violence between family members. More than one quarter of participants experienced some form of childhood sexual abuse. The chi-square analysis showed that females experienced more sexual abuse than males. Witnessing family violence and experiencing childhood physical abuse were consistent across genders.
Chi-Square Tests of Trauma Variables and t Tests for Differences in Means of Social Network Variables.
p < .001.
Table 2 presents t-test results for equality of means of the social network variables compared by gender. Independent sample t tests revealed few differences in social network variables between the groups. There was a trend (p < .10) in the difference of the number of male friends. Females appeared to have slightly fewer male friends in their network than their male counterparts. Other than this, there are no significant differences in the social network variables tested for males and females.
The four sets of OLS regression models stratified by gender are presented in Table 3. Model 1 revealed that females who experienced IPV in the past year had significantly more friends. Childhood trauma was related to females’ overall number of friends. For female participants, those who witnessed family violence during childhood had more friends, but those who experienced sexual abuse had significantly fewer friends. On the contrary, IPV experience was not significantly related to males’ overall number of friends. Unrelated to IPV, for males being older was significantly related to having fewer friends and time homeless was related to having more friends.
Multivariate OLS Regressions of Total Friends Stratified by Gender.
p < .05. **p < .01. ***p < .001.
The number of street-based friends and home-based friends also presented different results across genders. As seen in Model 2, older females identified fewer street-based friends. Males with a longer experience of homelessness have more street-based friends. Heterosexual females have fewer street-based friends than non-heterosexual females. The only significant result for both males and females is that White youth had more street-based friends compared with their non-White counterparts.
There are many significant differences by gender in regard to home-based friends, as shown in Model 3. Older males identified fewer home-based friends. Females who have spent more time homeless and who experienced childhood sexual abuse have significantly fewer home-based friends.
The results of the OLS regressions looking at the number of male friends for male and female homeless youth follow a similar pattern as the regressions for the total number of friends. As depicted in Model 4, older males have significantly fewer male friends. Females who identify as White, who have experienced IPV, as well as those who witnessed family violence have significantly more male friends. In addition, females who experienced childhood sexual abuse have significantly fewer male friends.
Discussion
The present study demonstrates how IPV experience is related to a youth’s peer relationships within a large sample of homeless youth. Consistent with nationally representative samples of adolescents (Glass et al., 2003; Hickman, Jaycox, & Aronoff, 2004), approximately one quarter of the homeless youth in this sample have experienced violence inflicted by a boyfriend or girlfriend in the previous year. To our knowledge, no previous study has explored how IPV relates to a homeless youth’s peer relationships. Prevalence of IPV was not related to any demographic variables, including gender, race, age, or sexual orientation. However, with further examination of the sample stratified by gender, it is clear that the network-level associations with IPV are different for homeless female and male youth.
This study suggests that the experience of IPV has substantial associations with the homeless female youths’ friendship network types. The limited previous research regarding social networks and IPV indicate that IPV victims frequently have impaired networks of social support (Levendosky et al., 2004) and abuse severity can negatively impact the number of close friendships and the frequency of contact with these friends (Coohey, 2007). However, these findings are based solely on a sample of adult women commonly leaving marital relationships with shared children. In this current study, females who experienced IPV identified having more friends in their network than those who have not reported past year IPV. While this result seems to be contradictory to previous literature, it is important to understand that the context of youth intimate relationships is often very different than adult relationships. Youth relationships differ in many ways, including duration, level of commitment, amount of prior relationship experience, peer-like status of partners, and causes and resolutions of conflict (Furman & Wehner, 1997; Laursen & Collins, 1994; D. A. Wolfe et al., 2001).
Moreover, IPV was significantly related to female homeless youths’ number of male friendships. This finding suggests that there is something specific about a female homeless youth’s relationships with male peers that is specific to victimization. For females, having a social network that is dense with male street-based friends is associated with experiencing IPV. It is possible that being in a male-dominated social network may increase the likelihood of victimization for female homeless youth. This complements the previous finding from Casey and Beadnell (2010), which showed how the structure of male adolescent peer networks increased the risk of IPV perpetration. Young males in networks characterized by small, dense mostly male peer networks with higher levels of delinquent behavior reported higher rates of perpetration. It is possible that for females, being in a male-dense network increases the exposure to perpetrating males. It is also possible that male-dense networks are more prone to using violence as a conflict tactic.
This study also found that witnessing family violence was associated with an increased number of male friends in a female’s social network. Previous research has confirmed that witnessing family violence in childhood impacts psychosocial adjustment for adolescents (Fergusson & Horwood, 1998). Exposure to inter-parental violence increases adolescents’ risks of anxiety, conduct disorder, problems with alcohol, criminal offending (Fergusson & Horwood, 1998), aggression toward peers, and depression (McCloskey & Lichter, 2003). The findings from the current study add to this body of literature suggesting that witnessing violence in childhood may contribute to the process of homeless females entering male-dense social networks. How this is related to psychosocial development and adjustment should be further researched. Moreover, the current data indicate that females who experienced childhood sexual abuse reported fewer overall friendships and fewer male friendships. Thus, future research should investigate how personally experienced abuse is related to fewer friendships, while witnessing violence is related to more friendships.
Male homeless youth evidenced very different network associations than their female peers. Older male homeless youth reported fewer friends, suggesting that they may have a tendency to isolate themselves. On the contrary, males who have spent more time homeless tend to have more friends. These findings were not true for their female counterparts, which also reflect the adolescent peer relational styles presented earlier by Rose and Rudolph (2006).
There are several limitations for this study. First, the participants in the study were recruited through a non-probability sample; therefore, results could be affected by sampling bias and the results of the study might not be generalizable to the larger population of homeless youth. The data were also cross-sectional; thus, causality between these constructs is not clear. It is likely that poor-quality peer relationships may be both a consequence and an antecedent of IPV for homeless youth. The IPV measure captured an individual’s experience in the previous 12 months, while the social network measures were related to the previous month. This difference in time frame contributes to uncertainty in the conclusions because it is possible that the measure of youths’ current social network does not reflect their network at the time of abuse. Currently the research team is investigating how the variability in a homeless youth’s social network over time. It is hypothesized that overall there is minimal change in a youth’s network over time with the exception of youth that experience notable life changes such as entering housing or adopting sobriety. Finally, the measure used to assess IPV was limited. It did not take into account other forms of IPV, including coercion, and emotional, mental, verbal, or sexual abuse. It is possible that many individuals involved in violent intimate relationships may not identify or disclose it through this measure. However, the measure was selected because it was a recognized scale included in the Youth Behavior Risk Survey, so as to allow for comparison with research regarding housed youth.
This study provides several directions for future research. First, given the prevalence of IPV within this sample, more information regarding the nature of the IPV experienced including severity and frequency is necessary. Mixed methods research designs that include qualitative data could provide this insight. Second, future studies should be designed to assess for a wider range of the different forms of IPV, as well as to assess for intimate violence perpetration. In fact, this research team has expanded to include the Revised Conflict Tactics Scale (Straus, Hamby, Boney-McCoy, & Sugarman, 1996) as a measure of IPV victimization and perpetration in their current study to further examine this phenomenon. Third, given that the results show that a homeless youth’s social network structure is associated with experiences of IPV, research should continue to be done to see how social networks might act as potential buffers to health and mental health outcomes. Finally, longitudinal data are necessary to make causal inferences of how IPV and social networks are related.
Overall, this study has implications for future practices. This study confirms that IPV experienced by homeless youth is an issue that needs to be addressed. More than one fifth of this sample identified experiencing violence inflicted by an intimate partner in the past 12 months. Previous research has shown that network embeddedness and ties to street-based peers may have negative effects on outcomes such as drug use and sexual risk behaviors (Rice, 2010; Rice et al., 2012), yet these results provide a different perspective. Knowing how friendship relates to IPV is important when designing and adapting interventions targeting this population.
Homeless youth function in a different social context than the general youth population. Youth immersed in street life have less access to adult role models such as parents or school-based employees. Even though many receive services at drop-in centers and frequently access caseworkers, their primary source of social connections and influence are their street-based peers. Therefore, peer-based interventions may be the most effective for this population. IPV interventions should be designed to facilitate healthy connections between homeless youth emphasizing building healthy relationship skills and non-violent forms of conflict resolution. Furthermore, considering the results regarding the differential impact on gender, gender-specific programs and interventions should be developed.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.