Abstract
Neighborhood context is recognized as an important predictor of individual-level behaviors and health outcomes (Pickett and Pearl 2001; Lee and Cubbin 2002; Sampson 2003). Neighborhoods, however, are difficult to define both in theory and in practice, and are often drawn to follow existing administrative boundaries or sampling schemes, or must be set arbitrarily due to a lack of sufficient data. Given the role of neighborhood context in influencing health outcomes, it is crucial that the area of influence surrounding the unit of analysis (be it a person, household, etc.) be properly defined. As already discussed in previous chapters, if we do not identify neighborhoods correctly, we cannot properly evaluate neighborhood effects. Defining neighborhoods is a challenge across the social sciences; investigations of the role of neighborhood context in decision making and shaping of individual-level outcomes are seen in public health, geography, demography, and sociology with no consistent approach to identifying and evaluating neighborhood effects. In this chapter, we outline several types of neighborhood definitions from the literature, and then, using data from the Women’s Health Study of Accra, implement a spatial model together with a simulation approach to examine how two alternative neighborhood definitions affect modeling of individual-level health outcomes.