Accuracy of Homeless Youth's Perceptions of Drug Use Among Their Peers: A Multi Level Analyses

Background: Homeless youth engage in a higher level of substance-use compared to their housed counterparts. Social network analysis is an appropriate method for understanding the peer context of adolescent substance-use. Studies using social network techniques to study drug-use among homeless youth have typically found that peer drug use is one of the most consistent predictors of their own use. The majority of studies examining peer influence on drug-use among this population have however relied on perceptions of peer use than actual friend’s reports. The validity of research that relies on adolescents’ perceptions of their friends’ substance-use behavior has been called into question due to the potential for rater bias, whereby respondents misperceive the behavior of their peers as a function of their own behavior (known as the false-consensus affect).  Differentiating the effects of actual peer use from perceptions of peer use may not only help us understand the optimal approach to studying peer effects on substance-use but also help guide future intervention efforts. Because network partners were also interviewed in this study, we are able to compare the respondents’ reports of their alters’ behavior with these alters’ self-reports.  In addition, sociometric-methods provided the opportunity to assess other network characteristics that may bear on the relation between peer and individual substance-use, such as network density, centrality of the individual in the network, and network position.

Methods: Event Based Approach (EBA) was used to delineate the boundary of the sociometric-network of homeless-youth, who were all accessing services at two drop-in centers in Los-Angeles, CA and Santa-Monica, CA (n=377)  between 2011 and 2012. Hierarchical-linear-modeling was utilized to investigate participant-level, network structural-level, and relationship-level factors with youths’ accuracy or inaccuracy of their perceptions.

Results: The accuracy of perceptions was high (between 70-90%depending on the substance). We found a strong relationship between the reported behavior of egos and their reports of their alters’ behavior for injection drug use, marijuana, methamphetamine, cocaine, heroin, and prescription drug use (OR=45.68, 95% CI=16.54-126.12; OR=37.09, 95% CI=12.01-114.53; OR=17.98, 95%CI=8.72-37.07; OR=47.49, 95% CI=19.09-118.15; OR=625.97, 95% CI=77.40-5062.32; OR=7.95, 95%CI=4.935-12.833, respectively).  The individual and network level factors associated with accuracy or inaccuracy varied by substance. Homophile in gender was associated with better accuracy about methamphetamine-use (OR=1.71, 95% CI=1.08-2.71). Network members’ out-degree centrality was associated with less accuracy about methamphetamine-use (OR=0.84; 95% CI=0.75-0.93). Participants occupying more central position in the networks were more likely to respond accurately about their alters’ heroin use (OR=1.58, 95% CI=1.14-2.19; OR=1.31, 95% CI=1.05-1.64).

Conclusion: These results suggest that there is a high level of accuracy in youth’s perceptions of their peers’ behavior. The findings also suggest that these perceptions tend to be more accurate if they are using the same substance. Also, there are relationship and network level factors that affect the accuracy of these perceptions. Furthermore, these results have implications for intervention design. Interventions designed to change norms have tended to assume that perceptions about drug use-are inaccurate and target these misperceptions. This might not necessarily be the most effective strategy in the light of these results.