Methods for identifying and assessing groups in health behavioral research.

Methods to study groups developed in other fields have much to contribute to the study of health behavior, including alcohol and substance use. This paper surveys group methods in order to demonstrate how attention to group levels of analysis can provide new insights about the context and causes of health behavior. The first major section of this paper focuses on the influences on health behavior associated with "naturally occurring groups" such as peer networks, families, workplaces, neighborhoods, churches, treatment groups and residential settings. Different aspects of group influence are discussed in terms of seven different features of social groups: history, structure, function, resources, process, norms and climate. The second major part of this paper is concerned with methods to identify meaningful subgroups in heterogeneous samples. Following an overview of taxonomic methods to identify and assign cases to groups, we highlight the potential for these methods to help address a number of fundamental theoretical problems in health behavioral research, including: describing cultural diversity, distinguishing response sets in survey research, controlling error in outcomes scores related to baseline differences, tracking trajectories of change, and revealing interactions among determinants of health behavior. The third and final section of the paper highlights specific applications of taxonomic and group-influence methodologies in intervention research, including design and analysis of randomized trials, studies of existing treatment settings and group-level variables involved in translational research.