Nov 1, 2011, 12:00 PM, Posted by Greg Duncan
By Greg Duncan, PhD, and Jens Ludwig, PhD, co-winners of a 2009 Robert Wood Johnson Foundation Investigator Award in Health Policy Research.
At different points in time, each of us has spent a blissful sabbatical year at the Russell Sage Foundation and living on the Upper East Side of Manhattan – just a few blocks from Central Park to the west and Weill Cornell Medical College to the east, with a Food Emporium right in the basement of our apartment building. Among the other striking things about the Upper East Side is how healthy people are – only 8.4 percent of residents were obese in 2003-07, the lowest rate in all of New York City. Yet just a five or ten minute ride north on the 6 train takes you to East Harlem, where nearly 30 percent of residents are obese (Black and Macinko, 2010).
These sorts of massive disparities across neighborhoods in health outcomes have generated long-standing concern that living in a disadvantaged neighborhood environment might causally contribute to adverse health outcomes, and so doubly-disadvantage poor families who are already at elevated risk for adverse health due to their own low incomes. Common hypotheses for why neighborhood of residence might contribute to obesity and closely related health problems such as diabetes include differential access to grocery stores that sell healthy foods, opportunities for physical activity, or medical treatment. Neighborhoods could also systematically differ with respect to social norms around health-related behaviors, or in terms of levels of psychological stress due to differences across areas in rates of crime and violence.
Empirically isolating the causal effects of neighborhood environments on health has been challenging for social scientists and medical researchers because most families have at least some degree of choice over where they live. Suppose we observe two observationally equivalent people, one living in a distressed area and the other in a more affluent area, with different health outcomes. Is the observed difference in health due to something about the neighborhood environments in which the two people are living, or instead to hard-to-measure characteristics of the two people that are related to their residential choices and directly related to health as well?