within a measured geographic area assessed on a per capita and/or a per land area basis, with healthy foods associated with grocery store and supermarket availability and less-healthy foods associated with convenience store and fast-food restaurant availability. Availability gets at the time costs associated with food shopping (e.g., a convenience store that is a five-minute walk away versus a supermarket that is a half-hour bus ride away), whereas affordability is the monetary cost or purchase price of various items. Based on their available price data, the affordability of healthy foods is represented by the prices of fruits and vegetables and of less-healthy foods by the prices of fast food and soft drinks. Taken together, Powell mentioned that availability and affordability determine the total cost of food, or its accessibility.

To provide an overview of accessibility, Powell and her team use data from the American Chamber of Commerce Researchers Association (ACCRA) for food prices and from Dun and Bradstreet (D&B) for outlet density. Drawing on D&B data, supermarkets and grocery stores are distinguished from convenience stores by the assumption that access to a convenience store alone does not provide access to quality food. Supermarkets are substantially larger food stores than grocery stores and are more likely to have onsite food preparation such as a butcher, a baker, and a deli. Chain stores are studied because they often benefit from economies of scale in terms of purchasing power, distribution, and other factors that contribute to lower prices. A validation study is under way to ensure that the outlet data available from D&B and infoUSA, another proprietary business database, do not contain biases across neighborhoods of different socioeconomic status and racial or ethnic characteristics.

Using D&B data, 29 percent of zip codes nationwide do not have a grocery store or supermarket, and 74 percent do not have a chain supermarket. Powell stressed that using zip codes alone is misleading, given that some zip codes contain no or very few people, and therefore she narrowed in on more densely populated urban areas. Of these urban areas, 7 percent have no grocery store or supermarket and 53 percent do not have a chain supermarket.

When her team looked at food availability by linking D&B data in 28,050 zip codes with U.S. Census data on race, ethnicity, income, population, and degree of urbanization for the year 2000 (Powell et al., 2007), based on multivariate models, quite significant differences emerged:

• African-American populations had half as much access to chain supermarkets as Caucasians, controlling for other factors;