Drifting Smoke Rings: Social Network Analysis and Markov Processes in a Longitudinal Study of Friendship Groups and Risk-taking 1

Abstract

Social network analysis is applied to three time points of a longitudinal study, which examines how risk-taking (represented by smoking and cannabis use) in adolescence is associated with social position within peer group structures. One hundred and fifty two students in the second year of secondary education in one Scottish school named up to six best friends, allowing for the categorization of each adolescent as a group member, a group peripheral or a relative isolate. Building on previous work, it is shown that transitions from non risk-taking to risk-taking behavior occur predominantly at peer group, rather than peripheral or isolate membership level. The transitions of pupils from time point one through to time point three are modeled as a Markov process, based on the assumption that the social position and risk-taking behavior (or transitional state) of a pupil at a certain time point depends only on their state at the previous time point. The results show that the Markov process is not stationary. The expected length of time spent by pupils in the various transitional states is also modeled, and provides another (time) dimension to the influence of peers on risk-taking behavior. We hypothesize that the influence exercised by an individual in a social network context increases with the cohesiveness of the individual’s social network position and the length of time he or she occupies that position. The results testify to the importance of risk-taking peer groups, both as a source of influence and selection of peripheral members, and to the need for differential targeting by sociometric state when delivering health education and intervention programs.