Understanding the social determinants of behaviours: can new methods help?

An individual’s behaviourundoubtedly results fromasetof comlicated processes involving interactions between environmental actors, personal characteristics, and biology. Moreover it is clear hat social processes, both through their influence on the social ndphysical environments inwhichpeople live andwork, aswell as hrough the transmissionofnormsandattitudes throughsocial netorks, play a key role in shaping behaviours. Understanding these ocial processes and theways inwhich they affect behaviour is funamental to the identification of the most effective interventions o improve health and reduce inequalities in health. The papers by alea,Hall, &Kaplan (this issue) andCooper, Bossak, Tempalski, Des arlais, and Friedman (this issue) emphasize the need to study these ocial determinants in the specific case of drug use. More specifially, they discuss important methodological challenges involved nd suggest and illustrate the application of new methods. Cooper t al. focus on the measurement of spatially defined environments ndGalea et al. focusonuseof agent-basedmodels tobetter capture ow relations between individuals affect population-level patterns f drug use. A first step in studying the social determinants of health ehaviours is the development of theories regarding the relevant ocial factors and the ways in which they may operate. Although a pecific methodology is not a prerequisite for the development of heory, the use of a specific analytical approach can influence the ypes of theories that are developedbecause the analyticalmethods eterminewhat can be tested andwhat cannot. For this reason, the dvent of new methods may sometimes stimulate new theories. or example, the advent of multilevel analysis stimulated much eeded theorizing on the ways in which group-level or contextual actors may affect health. The geographical and spatial methods llustrated by Cooper et al. may stimulate more sophisticated theoizing on the ways in which space shapes the distribution of health elated outcomes. The dynamic agentmodels illustrated byGalea et l. may stimulate theorizing on theways inwhich dynamic interacions between people, and between people and their environments ffect behaviours and health outcomes. One of the challenges to pidemiologic research today is that the excessive reliance on a