Examining the Geography of Food Deserts and Food Swamps in Austin,Texas
MetadataShow full metadata
In the past two decades, retail food environment exerts a tremendous fascination with scholars because it can shape individuals' eating behaviors and health outcomes. Food insecurity has been emerging as a priority for many food environment studies. Food deserts (defined as limited access to healthy foods) and food swamps (defined as overexposed to unhealthy foods) are two critical components of food insecurity.
Despite a lot of progress has been made in food environment studies, current retail food environment assessment mainly uses simply descriptive food assess measures, mostly overlooking the role of multiple transportation modes in food access, spatial associations between geographic food access and sociodemographic deprivation, as well as the variations of in-store characteristics across different food stores. This dissertation seeks to fill up the research gaps through pursuing three research objectives.
First, taking advantage of Geographic Information Science, neighborhood community nutrition environment was characterized using a proposed multiple-mode Huff-based 2SFCA method to measure geographic access to food outlets in Austin, Texas. The spatial accessibility score was calculated with a set of impedance coefficients ranging from 1.2 to 2.2. It shows an urban-core and peripheral disparity in terms of spatial accessibility; the spatial access to both healthy and unhealthy food outlets increase as the impedance coefficient increases. The proposed multiple-mode Huff-based 2SFCA was compared with its single-mode counterpart using a paired t-test. The comparison illustrates that the difference between the two methods on healthy food access was significant at the impedance coefficient 1.4; for the difference in unhealthy food access, it was significant at the impedance coefficient 1.5.
Second, this dissertation research examined the relationship between geographic accessibility to food outlets and sociodemographic marginalization at the block group level. Eight sociodemographic deprivation variables were reduced to two indices: the Economic Deprivation Index (EDI) and Sociocultural Deprivation Index (SDI). Different from the research that uses conventional statistics, this dissertation used spatial statistics to adjust for spatial autocorrelation and spatial heterogeneity problems in the relationships between the two entities. I first employed the spatial autoregressive model to account for the spatial autocorrelation issue. The spatial lag model shows that block groups' EDI was significantly related to the access to healthy food (coefficient = -0.054, p = 0.037); spatial error model shows that SDI was significantly associated with the access to unhealthy food outlets (p = 0.000). This finding indicates that block groups in low economic deprivation (representing high economic status) enjoyed better spatial access to healthy foods, while those in high sociocultural deprivation (indicating low sociocultural status) were overexposed to unhealthy foods. I then used a semi-parametric Geographic Weighted Regression (GWR) model to explore spatial heterogeneity in the relationship between spatial food access, EDI, and SDI. The semi-parametric GWR allows the flexibility to incorporate both fixed and geographically varying explanatory variables, providing a more satisfactory model fit than the conventional GWR. The result shows that the EDI was a significant global predictor of healthy food access (p = 0.000) but was an insignificantly global predictor of unhealthy food access (p = 0.061); the SDI is a varying local variable to predict both healthy and unhealthy food access. Also, the spatial access to food outlets, EDI, and SDI were integrated to identify food deserts and food swamps in Austin. The use of hot spot analysis enables me to account for the spatial dependence issue that was ignored by previous studies. The result shows that food desert neighborhoods were mainly located in the eastern part of IH-35, and food swamp neighborhoods were in the northeast corner of Austin.
Finally, this research fills up a research gap that lacks studies examining the consumer nutrition environment in food-insecure (e.g., food desert and food swamp) and food-secure (e.g., food oasis) neighborhoods. It investigated consumer nutrition environment (i.e., food availability, food price, food quality, and labeling) in three selected neighborhoods (e.g., food desert, food swamp, and food oasis) in Austin, Texas. A food auditing instrument m-TxNEA-S was developed in this dissertation to capture the unique dietary culture and food preferences for Hispanic/Latino groups in Texas. Then this surveying tool was used to survey 14 grocery stores and 32 convenience stores in the three neighborhoods. It shows high inter-rater reliability (mean = 0.96) of the m-TxNEA-S. The result shows that there was a statistically significant interaction (p = 0.019) between the effects of store type (ST) and neighborhood nutrition environment (NNE) on healthy food availability. Food swamp and food oasis neighborhoods’ grocery stores offered significantly more healthy foods than convenience stores. Grocery stores in the food swamp (p = 0.018) and food oasis neighborhoods (p = 0.015) had a significantly higher availability of healthy foods than those from the food desert neighborhood; convenience stores in the three neighborhoods did not exhibit any significant difference on healthy food availability (p = 0.932). It also shows that the prices for healthy items (lower fat, lower calorie, and whole grain) were not significantly different from their regular food options by ST (p = 0.374) and NNE (p = 0.437). The analysis of food quality shows that there was not any significant difference by ST (p = 0.801) and NNE (p = 0.272). It did not exhibit any significant difference in food labelling by ST (p = 0.897) and NNE (p = 0.802).
CitationJin, H. (2019). Examining the geography of food deserts and food swamps in Austin,Texas (Unpublished dissertation). Texas State University, San Marcos, Texas.