Abstract
Finding Food Deserts: Methodology and Measurement of Food Access in Portland, Oregon
Andrea Sparks, U.S. Department of Housing and Urban Development; Neil Bania and Laura Leete, University of Oregon
Introduction
(Please see complete paper for list of references)
In recent years, a growing understanding of the linkages between diet and health has led
to increased scrutiny of the accessibility of a wide-range of competitively priced healthful foods
in urban environments. Studies in the U.S. first identified ‘grocery store gaps’, inner-city areas
experiencing disinvestment in retail grocery stores leaving low-income inner-city areas
underserved by traditional grocery store retailing. This was followed by a number of U.K.
studies that further refined the questions and research methodologies for defining ‘food deserts’– low income, urban areas with diminished walking distance access to grocery stores. Most recently, a number of
studies have asked similar questions for Canadian cities. A range of patterns have emerged, from
findings of pronounced food deserts in some locales (e.g. London, Ontario) to findings of a
relatively even distribution of grocery store access in others (Montreal, Quebec and Edmonton,
Alberta).
To date, the same research methodology for examining food access has not been as
widely applied to U.S. cities. Cotterill and Franklin (1995) documented a statistically significant
and negative relationship between public assistance rates, lack of car ownership and square feet
of supermarket retail space at the zip code level in 21 major urban areas. Morton and Blanchard
(2007) produced U.S. wide county-based measures of access to supermarkets. Zenk et al. (2005)
relate disparities in distance to supermarkets to poverty and demographic characteristics in Detroit, Michigan. However, to our knowledge, to date no researchers have published a city-
specific GIS based analysis of food deserts located within a U.S. metropolitan area. In this
paper, we seek to fill that gap and conduct an analysis of food deserts and related demographic
and methodological issues for the Portland, Oregon metropolitan area.
This study has three major components. First, we follow the methodology employed in recent papers on Canadian cities (Smoyer-Tomic et al., 2006, Apparicio et al., 2007, and Larsen and Gilliland, 2008) to investigate whether the Portland metropolitan area has low-income areas lacking access to supermarkets that one would consider to be ‘food deserts’. Second, we investigate whether there is more generally a pattern of unequal supermarket access between higher and lower-income areas, or across areas with different demographic characteristics. Finally, we use the data from the Portland area to examine a number of questions regarding the sensitivity of food access measures to methodological variants. In particular, we are interested in determining whether more and less computationally intensive methodologies yield consistent results.
This latter point is particularly important vis-à-vis the ability of practitioners to compute and use accessibility measures for local planning. Less computationally intensive approaches to food and other access measures may be within reach of local planners and policymakers, while more computationally intensive measures may not be. In particular, we look at two methodological questions. First, we follow up on the investigation by Hewko et al. (2002) of aggregation errors in measuring accessibility to a number of urban park and recreation amenities. We also examine differences in measures that result from using either network distances or the more easily computed Euclidean distance. As Pothukucki and Kaufman (2000) have noted, planners have not systematically considered food provisions systems or food access questions. Less computationally expensive approaches to measuring accessibility in urban areas could help remedy this situation.
Download final conference paper
