Most research problems in epidemiology are multifaceted and, therefore, complex. The fact that they are complex should not mean that the problems cannot be researched. Statistical methodology is responding to the complexity of the research challenges. In Merlo's invited commentary (1) on our article (2), he argues that although risk factor epidemiology can address some research questions, it is time to advance eco-epidemiology using appropriate statistical methods. Multiple membership multiple classification (MMMC) multilevel models have been around for some time (3) but are underused in social epidemiology. Multilevel models allow for individual factors (e.g., biological and lifestyle factors) and clustered groupings (e.g., neighborhoods, schools) to be studied at the same time. This enables investigation of the extent to which the different levels interact with each other or act independently of each other (4). An understanding of the factors at both individual and cluster levels and their relationships means that appropriate policy recommendations can be made (5).
展开▼
