Modeling Drug Use Using Empirical Social Networks

Social networks play a prominent role in a person's initiation of drug use, continuance of use, as well as cessation or resistance to initiation. And, each stage of drug use can correspond to different kinds of personal networks. For example, a person with close ties to family members and non-using friends may never try an illegal drug, while someone, whose social circle is composed of mainly drug using peers, is going to be subject to pressure and frequent opportunities for engaging in use [Kandel 1978; Kandel and Reveis 1989; Latkin et al 1995]. Also, stopping use and successful treatment are possible often with the help of a supporting network of friends and family [Gainey et al 1995; Ellickson and Bell 1990].

However, much of the research, that explores these network effects, has focused on just the local networks, or ego-networks. These networks consist of each subject's close contacts, mainly friends and family members, and sometimes their relationships to one another and, furthermore, lack the social structures beyond which would be the social environment of the subjects' contacts. We know little about the people who influence the people who influence the subjects, and this severely handicaps network analyses.

Also, this body of research has only recently begun to address policy concerns and make few concrete statements about the effects of network structure on interventions aimed at use, and ultimately harm, reduction; these include prevention programs (for those at risk of initiating), treatment (for addicted users), and enforcement (in the form of arrests). Instead, they primarily test theories of deviance and criminal behavior and describe the general sociological indicators of drug use, which, in and of themselves, are valuable pieces of information.

This proposal defines a research agenda that endeavors to fill some or all of the aforementioned gaps. I want to examine a complete, network structure, composed of both users and non-users alike, one that is whole in contrast to the disconnected ego-networks of prior research. I will include this network in a dynamic model of interaction and influence in order to test interventions, with the goals of isolating the structural conditions that contribute towards or hinders use reduction and revealing ways in which networks could be changed to assist the interventions and formulate network-based intervention strategies. Ultimately, I want to shed some light on how the embeddedness of the personal network within the community shapes and influences the drug-prone elements of the network and their use patterns.