A Path Model to Understand the Predictors of Gang-Involvement for Homeless Youth

Background: Homeless youth (HY) is a large and growing population with an estimated 1.6 million in America. Additionally, youth gang membership and related negative outcomes continues to be a national problem. These two social issues intersect as many HY identify as being a part of a gang or closely connected to a gang member. For this unique population little is known about what characteristics- individual and environmental- predict gang involvement in the homeless youth population. Using Thornberry, et al.’s (2003) interactional model of gang membership, the current study utilizes a path model analysis to determine what the salient predictors of gang membership are for homeless youth.

Method: Thornberry and colleague’s (2003) model adopts a life course perspective to understand predictors of youth gang involvement and delinquency. The model identifies various domains of risk factors that are related to youth joining a gang including: community, family, school, peers, individual and problem behaviors. A sample of 614 Los Angeles area drop-in service seeking HY completed a self-administered questionnaire. Youth were asked if they have ever or if the currently identify as a gang member.  Variables included in the model were community violence, experience of childhood maltreatment (witness of family violence, sexual abuse and physical abuse), and age of onset of risk and problem behaviors including sexual intercourse, alcohol and marijuana use and age of first involvement in juvenile justice system. To explore which of these variables were related to gang membership a recursive path model was created using the CALIS Procedure in SAS 9.4. The model takes into account both the direct effects of the early life and community variables as well as their indirect effects through age of onset of: sexual intercourse, drug and alcohol use and juvenile justice involvement.

Results: Fifteen percent the sample identified as gang current or former gang members. Results from the path analysis revealed that community violence is directly related to all adolescent risk behavior variables i.e. earlier age of sexual onset, alcohol and marijuana use and juvenile justice involvement. The experience of physical abuse was related to early age of sexual onset. The path model revealed no indirect paths, however community violence was directly related to gang membership.

Discussion: Homeless youth report high rates of a history gang membership; higher than what is found in housed youth populations. Results from the path analysis reveal that, for homeless youth, community violence is the most significant predictor of gang membership. This has several implications. For homeless youth, a population that experiences high rates of trauma, abuse and early engagement of risk behaviors, the exposure of community violence is the primary distinguisher of youths that have ever been involved in a gang. These results may reflect the impact of environmental level variables such as neighborhood and geographic location. Future studies should incorporate geographic level data. Further, communities with high levels of gang violence should consider youths’ risk for homelessness. The results have many implications for future gang and youth homelessness prevention practices.