The Importance of Marijuan Use to the Construction of Homeless Youth Networks

Objectives:  For decades, research has demonstrated that vulnerable adolescents engage in risk taking with their peers. Social Work has become increasingly interested social network processes  which extend beyond simple dyadic relationships into larger webs of interconnected individuals impact the well-being of persons who are a part of these vulnerable populations.   This so-called “sociometric” (or whole network) data rarely exist among populations such as homeless youth (HY) despite the overwhelming evidence that HY report more problem behaviors such substance use and abuse than housed youth. The transience of such populations increases the complexity of this research.   We examine “heavy” marijuana use (at least daily) among HY in population-level network data over time, hypothesizing that “heavy” marijuana use is a behavior around which social ties are forming, despite the transience of HY networks over time.

Methods:   Network data were collected from two populations of HY recruited from drop-in centers in Los Angeles, every six months for one year (site 1: n= 237, 263, 312; site 2: n=138, 149, 131). For each wave, a sociomatrix was generated based on youth nominating other youth in the sample.  K-core defined peripheral/non-peripheral network positions and degree centrality defined network prominence; logistic regression assessed associations with heavy marijuana use.

Results: Approximately 60% of the membership of network 1 and 85% of the membership of network 2 changed between each successive panel. For site 1, the association between network measures and heavy marijuana use varied over time, but with significant odds ratios (OR) for directed out degree (OR=1.21 at wave 1, OR=1.17 at wave 3), undirected degree (OR=1.22 at wave 1, OR=1.10 at wave 2), and kcore (OR=1.40 at wave 1, OR=1.24 at wave 2). Thus participants are more likely to engage in heavy marijuana use as individuals nominate more alters, as individuals are connected to more alters, and as general network density increases. For site 2, the impact of network measures on heavy marijuana use vary much more widely over time, with significant odds ratios for waves 1 and 2 for directed out degree (OR=1.22 for wave1, OR=1.35 for wave 2) and undirected degree (OR=1.17 for wave 1, OR=1.20 for wave 2), and significant odds ratios for waves 2 and 3 for low kcore (OR=.374 for wave 2, OR=.472 for wave 3).  For site 2, participants in waves 1 and 2 are more likely to engage in heavy marijuana use as individuals nominate more alters and as individuals are connected to more alters, while participants in waves 2 and 3 are more likely to engage in heavy marijuana use when networks are less dense.

Conclusions: Heavy marijuana use appears to be important to the construction and reconstruction of these networks overtime.  This research has implications for theory and practice.  Many interventions are based on social influence models, yet these data suggest that social selection around marijuana use is a powerful force.  New, potentially “disruptive” network interventions which do not normalize heavy marijuana use appear to be needed.