Defining Neighborhood Boundaries for Urban Health Research: A Case Study of Accra, Ghana

The neighborhood has been used as the unit of analysis for a variety of studies including health, wealth, and ethnic diversity (Yen and Kaplan 1998; Diez Roux 2001; Morland et al. 2002; Weiss et al. 2007; Agyei-Mensah and Owusu 2009; Santos et al. 2010). While this unit of analysis is acceptable to many social science researchers, defining what a neighborhood is has been left up to the individual researcher. In many cases a proxy such as the census tract, ZIP Code or other pre-defined unit of analysis is used. Using a pre-defined unit of analysis reduces the amount of work involved and allows for comparisons to be made between the phenomenon under study (i.e., health, wealth, race) and the information for which the unit was created (i.e., census demographic variables). While allowing for a more straightforward, simpler study, predefined units do not always represent socially meaningful areas that are of symbolic significance to residents or the actual neighborhood boundaries that residents would use to describe their neighborhood. In addition, the choice of the neighborhood unit can affect the results of any study because of the modified areal unit problem (MAUP) (Openshaw and Taylor 1979). Therefore, choosing the appropriate neighborhood boundaries is an important part of any neighborhood study.