The present study was a secondary data analysis with data collected in the STORE (STaple foods ORdinance Evaluation) study. The STORE study tested compliance to the ordinance among convenience stores and small non-traditional food stores (e.g., gas food marts, dollar stores), as well as impact on customer purchasing and home food environments.
Retailer compliance was assessed at four different time points during the implementation and follow-up period. Data were collected pre-policy (July – December 2014, hereafter called time 1), during an implementation-only phase (no enforcement; September – October 2015, hereafter called time 2), at the initiation of enforcement (May – July 2016, hereafter time 3), and after continued monitoring (August – December 2017, hereafter time 4). The STORE Study compared changes in food availability at stores between neighboring cities, Minneapolis (ordinance implementation) and Saint Paul (no ordinance). Ninety stores per city were randomly selected from a government list of stores with grocery licenses, excluding supermarkets, WIC-authorized, invalid licensing addresses or were exempt from the ordinance (n = 255). Following field visits, verification and consent, 159 stores actively consented to participate in the study at one or more of the four data collection points (21).
Store Assessments
Trained staff assessed the availability and price of 69 food items using a modified instrument from the Rudd Center for Food Policy and Obesity (26). The instrument lists items in specific package sizes for which availability, price and quality (e.g. fruits and vegetables) is recorded. The adapted instrument can be found here: https://conservancy.umn.edu/handle/11299/20378 (21). The parent study evaluated the change in the availability and price of all 69 items. Other store characteristics were collected including store ownership (independent versus corporate), SNAP authorization and store location (27). Store ownership may be an important determinant of manager decisions around stocking healthy food.
For the present study, culturally preferred foods were identified for this analysis. Culturally preferred foods from the STORE assessment were selected based primarily on data from The Food Group, an equity-focused local food bank that complied the food list through client requests and key informant interviews in its process of creating a cultural equity toolkit (10). This resource was selected because it provided a local perspective of high-demand foods informed by community leaders, and it was supplemented by published literature that included studies in other areas of the U.S. Notably, if a food appeared on the culturally preferred list, it did not imply that the food is uniquely appealing to or consumed by certain groups, only that the food may have demand, serve as a household staple, or be more commonly part of the life experience among certain groups. Along with other universal staple foods, culturally preferred foods contribute to a healthy diet.
For Black/African Americans in particular, the history of slavery heavily influenced identified culturally preferred foods (28). When slave ships kidnapped Africans for slavery, ships brought some crops from West African (e.g. watermelon, okra, peanuts) to the U.S. for the slaves to farm and eat. Other traditional foods were not available to the slaves, such as African yams, and they adopted the sweet potato as a similar food item (28, 29). The roots of “Soul Food” is a product of using available food ingredients to maintain elements of original West African meals. (29). Accordingly, Black/African American preferred foods were identified as bananas, peaches, blackberries, blueberries, raspberries, tomatoes, collard greens, corn, kale, okra, turnip, yam, dry lentils/peas, cornmeal, fufu, and millet (1, 30). Many East African (a subpopulation of the Black/African American) consumption patterns and preferences overlap with Black/African American, including dry lentils/peas, tomatoes, millet and yams (1). Additional cultural foods for East African communities were identified as dry beans, corn, and teff. Latinx preferred foods were identified as bananas, pineapples, avocados, guavas, limes, mangos, papaya, tomatoes, acorn squash, peppers, plantains, yellow squash, zucchini squash, dry beans, dry lentils/peas, whole wheat tortillas, white corn tortillas, white flour tortillas, and cornmeal/masa (1, 10) Commonly consumed Asian foods were identified as tofu, bananas, oranges, peaches, limes, pears, broccoli, green/red cabbage, bok choy, and eggplant (31, 32).
The 5-year American Community Survey estimates (ACS,2009–2014) (33) were used to determine community demographics. A community was defined as the census tract where each store was located. We identified four types of communities of color for this paper, where 20% of the census tract population was either Black/African American, Latinx, Asian, and East African (34). East African communities were determined by a 20% or greater Black/African American population with an additional language spoken at home. Once these communities were indicated, local knowledge was used to confirm the identification.
Analysis methods
The analysis included stores (corner stores, gas marts, and dollar stores), located in one of the four types of communities of color described above and that were assessed at both the pre-ordinance time point (time 1) and the 12 months post-enforcement time point (time 4). For this analysis, only time 1 and 4 were analyzed to reduce the number of statistical tests, simplify the interpretation of results, and reflect the time during we the biggest change was expected to occur (i.e, time 1 to 4).
Store characteristics were summarized overall for Minneapolis (ordinance) and Saint Paul (control) stores, and by the four types of communities of color for Minneapolis stores using descriptive statistics. Two-sample t-tests or chi-square tests (or Fisher’s exact tests where any cell count < 5) were used to compare the community and store characteristics between the Saint Paul and Minneapolis stores. Statistical tests were not run to compare stores in each community of color because they were not mutually exclusive.
Descriptive statistics (frequencies and proportions) were calculated for the availability of the cultural foods at time 1 and time 4 (for foods present in at least one store in Minneapolis at either time point), overall and stratified by the community of color. McNemar’s exact tests were computed to test for statistically significant changes in the availability of each cultural food (as well as any cultural food) from time 1 to time 4.. A paired t-test was computed to test for a change in the average number of culturally available foods from time 1 to time 4 within each community of color.
To test for the effect of the policy on the availability of at least one cultural food in stores, generalized linear mixed models with a random intercept for store. The outcome was the availability of at least one cultural food (yes/no) within each community of color (where model converged and n ≥ 20) and for the full sample. A second set of regression models were computed with Minneapolis stores only, overall and stratified by each community of color (where model converged and n ≥ 10), testing for a change in the availability of any cultural food from pre ordinance to 12-months post ordinance enforcement, accounting for store ownership type (independent vs, corporate).
Regression models were only computed when the number of stores was at least 10 per city and the total number of stores in the model was at least 20. Due to a small number of stores (n = 2) in Asian communities in Minneapolis, computing regression models were not possible for Asian communities. Similarly, due to the small number stores in Latinx (n = 3) and East African (n = 5) communities in St. Paul, models for these communities of color were limited to Minneapolis only. SAS v.9.4 (SAS Institute Inc., Cary NC) was used for analysis. P-values < 0.05 were considered statistically significant, and 95% confidence intervals were provided where appropriate.