Identifying Demographic Variables that Affect Local Food Access
I, Kimberly Hill, conducted a statistical analysis to 1) better identify which demographic variables affect low access to healthy food and 2) determine if local food variables affect food access.
The initial research question was “which socioeconomic variables affect the existence of food deserts in the 31 counties in question?” using ordinary least squares (OLS) analysis. For the purposes of incorporating local food variables, I used a different Poisson model given that the only consistent variables for local food were count data.
All data used in this statistical analysis was sourced from The Food Environment Atlas, developed by the USDA ERS in 2012.. Based on the Food Environment Atlas, I selected the total percent of the population with low access to a major grocery store on the county level as the response variable for both the OLS regression and the Poisson regression.
Table 1. Results of Statistical Analysis
Results and Discussion
In comparison to a previous study using only the original 31-county study area in North Carolina, the results of this regression are very similar for the dummy variables (metro counties, population loss counties, and persistent poverty counties) but very different for the socioeconomic variables. For example, there was a positive correlation between % nonwhite and the response variable in the previous study, but this study indicates a negative correlation between these two variables. This may be due to omitted variables that differ from state to state and county-to-county. It may also be due to differences in ethnic makeup from county to county that are not accounted for in this model.
Ethnicity, persistent poverty (as represented by persistent poverty counties perpov00), population loss (as represented by counties experiencing population loss poploss00), and the percent of the population aged 65 and older as well as 18 and younger are key variables. The percentage of grocery stores per 1,000 residents is highly significant at the 10% level and is used as a control variable accounting for much of the variation in the percentage of a county’s population experiencing low access. However, poverty rate has some effect but is not significant in this model. This may be due to multi-collinearity with other variables, including % 65 and older, % 18 and younger, and ethnicity.
Age is an important factor in food access. This is most likely because seniors, who may live in isolated areas without access to transportation, and children, who are dependent upon their parents or guardians, are over- or under-represented in the response variable (% of population facing low access to grocery stores) due to their extreme dependence on those around them and geographic location. Higher percentages of the population younger than 18 or older than 65 are negatively associated with low access to grocery stores. The response variable may also be affected by programs to increase access for senior and minors, though I have no indication of this based on Food Environment Atlas documentation.
Applicability to North Carolina
It is unsurprising that a number of variables such as household income, ethnicity, and education are highly correlated with one another, just as there is, in fact, very little revelation in the fact that those who are poor or non-white have less access to quality, affordable food than other communities within North Carolina, given its long history of segregation. As is visible in the map below, food deserts are persistent throughout the state (ERS, 2013).
According to data recently released by the USDA, North Carolina currently ranks as the state with the fourth highest level of food insecurity in the United States, with 1 in every 6 North Carolina households reporting they struggled to provide or could not access affordable and nutritious food within the past year (Coleman-Jensen, Nord, and Singh, 2013). Given the results of this study, I can, with reasonable confidence, state that race, poverty and income are all significant variables affecting access to food—wealthier, non-minority communities consistently have better access to fresh affordable food products.
 The Food Environment Atlas is available online at http://www.ers.usda.gov/data-products/food-environment-atlas/about-the-atlas.aspx#.UpzLNsSsim4.