Neighborhood Effects on Health: Why Epidemiology Needs More Sociology
J. Michael Oaks (Division of Epidemiology and Community Health)
Identification and estimation of the independent effect of (neighborhood) contexts on behavior and health outcomes remains, as one distinguished sociologist put it, the Holy Grail of social science. Whereas most sociologists seem to have headed the early cautions and warning of methodologists such as Hauser, and Blalock and others, the emergence of the so-called multilevel regression model was seen by many outside sociology as a tool to overcome early concerns. This seems especially true in epidemiology, where it seems fair to say that multilevel models are 'all the rage', despite hard-hitting criticism from Oakes and a handful of economists such as Durlauf and Blume. Unlike sociologists who seem to appreciate much of the complexity involved, contemporary (social) epidemiologists appear remarkably unaware of key issues such as residential segregation, the culture of poverty, social stratification, and social interaction. This talk will address these issues from a methodological perspective, and show that if it is to be of any use, epidemiological inquiry into neighborhood contextual effects must incorporate sociological research and understanding.