Understanding Access to Grocery Stores in Food Deserts in Baltimore City


As American cities have seen a change in land uses in urban areas, with stores moving to suburban areas increasing the reliance on cars, many urban areas are left with a lack of accessible quality food options. These food insecure areas, commonly known as food deserts, where residents lack both access and sufficient economic resources to purchase healthy food, result in health disparities for residents in these communities. Though the existence of food deserts are well known among researchers, there is a lack of consensus on how food deserts are defined and identified. Through a survey of 573 Baltimore City residents, this study provides an in-depth analysis of individual grocery store choice and travel decisions. The study found that most people grocery shopped 2-4 times in a given month and at 2-3 different grocery stores; the choice of the store depends on the items purchased. In evaluating food desert metrics, two common assumptions are made: (1) trips originate from home and (2) people shop at the nearest store. This study found that the second assumption does not hold as an overwhelming percentage of those surveyed (77%) do not shop at their nearest grocery store

By using the survey data, the authors of this study identified user-generated data-driven indicators with statistical significance for developing a novel food desert metric using CHAID decision trees. A new healthy food priority area measure was developed for Baltimore that deemed all residential areas where the median income of the census block group is less than $35,000 as food insecure. A prioritization matrix was developed based on the secondary factors of proximity to the nearest grocery store (at the half-mile threshold) and the number of stores within 3 miles. This measure found a significant difference in the frequency of grocery store visits as well as the quality of food for those who live in a food desert as opposed to those who do not. The results of this study showed that people value choice of stores when grocery shopping. Limiting food desert measures to the distance to the nearest supermarket undervalues the importance of choice and variety in food selection. The data-driven yet simplistic methodology presented can be replicated to other municipalities as developing an accurate method of prioritizing areas for investments to reduce food disparities is vital to addressing the prevailing systemic divestiture of resources on communities.

Read the full report.

Impacts and Outcomes

The study provides a systematic, evidence-based methodology for determining the geographic areas which are food insecure through the analysis of individual choice survey data. Though transportation plays an important role in accessibility, the CHAID decision tree analysis of the accessibility indicators found that vehicle ownership was not a predictor of grocery store accessibility and that income was the primary factor.

Universities and Sponsoring Organizations Involved

US Department of Transportation Office of the Secretary-Research

Morgan State University

Principal Investigators

Dr. Celeste Chavis, email: celeste.chavis@morgan.edu

Anita Jones, M.S., email: anita.jones@morgan.edu

Funding Sources and Amounts

USDOT: $60,000

Completion Date

October 2020


Food desert, access, mode choice