Who shops at their nearest grocery store? A cross-sectional exploration of disparities in geographic food access among a low-income, racially/ethnically diverse cohort in central Texas

ABSTRACT

We examined whether Central Texans shop at their nearest supermarket, how far they travel for groceries, and explored differences by race/ethnicity, urbanicity, motivations for store selection, and other demographic characteristics. Using cross-sectional data and GIS, continuous network distances from participants’ homes to nearest and usual supermarkets were calculated and multivariate linear regression assessed differences. <19% shopped at their nearest supermarket. Regression models found that urbanicity played a large role in distance traveled to preferred supermarket, but other factors varied by race/ethnicity. Our findings demonstrate that racial/ethnic and urbanicity disparities in food access and multiple domains of food access need greater consideration.

KEYWORDS:

• Food Access
• Racial/ethnic disparities
• Geographic food access
• Geospatial analysis
• Urbanicity disparities

Introduction

Conceptualization of geographic access to food and disparities

Geographic access to food is defined as the physical presence of affordable, culturally relevant, and safe food within an individual’s community.^1–3 One of the most notable disparities in the geographic food access literature is by race/ethnicity.⁴ In multiple systematic reviews, scholars have found that areas with a predominant racial/ethnic minority population were found to be more likely to have limited geographic food access than predominantly non-Hispanic white communities.⁵ Specifically, black neighborhoods were much less likely to have a supermarket than white neighborhoods in multiple urban areas across the United States.^6–10 Additionally, Hispanic communities have also been shown to have less geographic access to healthy foods and were less likely to have culturally relevant healthy foods in their neighborhoods than non-Hispanic white communities.¹¹ Thus, communities of color are more likely to have limited geographic access to food retailers and less likely to have culturally relevant healthy foods available in their communities than predominantly non-Hispanic white communities.

Additionally, another noted disparity in the geographic access to food is by urbanicity. Urban/rural disparities are also frequently discussed in the literature.^12–15 Rural areas are often identified as having more limited geographic access to food due to less development and population density.^12,13,15 Studies exploring these disparities in rural settings are important,^14,16 however, communities are often not solely urban or rural, and the continuum of urbanicity needs greater consideration.^16–19

Geographic food access measures and limitations

The measurement of a complicated constructs such as geographic food access has employed varied and inconsistent methodologies. Geographic food access is most commonly analyzed in the literature using objective measures of food access. Objective measures of geographic food access include utilizing Geographic Information Systems (GIS) in order to calculate density or distance from an individual or population-centered centroid to food outlets.^17,20,21 However, using solely objective measures may not capture the nuance of individual and community behaviors, and thus fail to accurately understand the geographic food access issues experienced by residents in the area.^22–24 Also, much of the analyses of geographic food access only look at aggregated measures (density) rather than individual distances, which are needed in order to better understand barriers to geographic food access.^23,24 This is most often observed in the literature discussing “food deserts” and in the USDA ERS Food Environment Atlas that classifies individuals based on the density of food retail outlets in a zip code or census tract and centers on the assumption that people shop for groceries close to home.^23,25–30

Moreover, much of the food access literature does not take into account that individuals may not solely interact in the food environment near their neighborhood, or other geographic unit. This was emphasized in the work by Chen and Kwan (2015) which highlighted uncertain geographic context problem in food access research (UGCoP) in the food access literature.³¹ The authors critiqued that there are various issues with food access measurement, including that the current literature base assumes that an individual’s immediate food environment is centered around proximity to their home address, the lack of understanding of how an individual is interacting with their food environment in “real time,” the temporality of the food environment, and the perceived nutritional environment.³¹ Thus, greater research that utilizes objective measures paired with behavioral data is needed to understand how individuals interact with their food environment and to better understand the role of geographic food access.

Incorporating shopping behaviors and motivations into the geographic food access literature

As stated, much of the geographic food access literature solely explores the prevalence of food retail, without engaging with communities to understand individual behavior. This assumes that individuals are shopping at the stores located within their community, or at the stores that are closest to them.^32–35 The work by Hillier and colleagues (2011) found that this assumption is erroneous. In their sample of 198 low-income mothers in North Philadelphia, program participants rarely shopped at the supermarket closest to them, and traveled on average 1.58 miles for non-WIC shopping and 1.07 miles for WIC shopping.^33,36 This finding is particularly surprising since those average distances are over the standard metric used to measure limited geographic food access in urban areas.^20,24 This work has been further corroborated by Ver Ploeg and Wilde (2018), who conducted a literature review that concluded that grocery store geographic access as measured by distance modestly impacts shopping behaviors, but other household-level factors may play a more significant role.³⁷

However, solely incorporating individual shopping behaviors can create more questions, such as what motivating factors (or sociodemographic factors) could be driving store preferences and shopping behaviors. As highlighted by Wilde and colleagues (2021), there could be other aspects of the definition of access, such as affordability, availability, and cultural appropriateness, that could be shaping an individual’s food access and experiences in the food environment, but more exploration is needed.³⁸ Scholars have explored the roles of price, variety, quality, and convenient locations as drivers of food shopping behaviors, and specifically for bypassing food retail closest to home.^39–45 Most recently, the work by Shier and colleagues (2022) found that there was great heterogeneity in shopping behaviors among a racially/ethnically diverse sample in Los Angeles.⁴⁴ Shier and colleagues explored shopping behaviors as well as potential motivating factors for store selection among a low-income, urban sample, and found that distance traveled to preferred store varied by race/ethnicity and motivating factors (price, convenience, quality of food).⁴⁴ Work by Wei and collaborators (2022) similarly found that shopping behaviors were influenced by shopping motivations such as price and convenient proximity among a low-income sample in Atlanta.⁴⁵

Literature gaps

Therefore, examination of geographic food access should ideally include objective measures of distance to food retail, include data regarding an individual’s shopping behaviors and community-level factors, and be sensitive to disparities and intersectionality of these factors highlighted in the geographic food access literature. This need has also been highlighted by Singleton and colleagues (2020), who called for incorporating an intersectionality approach, and simultaneously socioeconomic status, race/ethnicity, and geographic location.⁴ However, there have been few studies that have explored geographic food access, shopping behaviors, and motivations, which have also taken an intersectional approach. Studies such as those by Shier (2022) and Wei (2022) and colleagues are exemplary, but are specific to lower-income urban communities, as are much of the other studies that have explored similar topics.^33,39–46 There has been some exploration of rural shopping behaviors, however, minimal along the range of urbanicity (i.e., communities that fall between urban and rural).^13–16,47 Thus, more work is needed that would employ a similar context-specific approach but in a region that has multiple communities along the urban-rural continuum. One such region that this could occur is Central Texas. The greater Austin area has experienced tremendous growth in the last decade while also having a racially/ethnically diverse population, and zip codes across the urbanicity spectrum.^48–50 Therefore, making the Central Texas region ripe for exploring relationships between geographic food access, shopping behaviors and motivations while adopting an intersectional approach.

Purpose of the study

The purpose of this study was to examine whether individuals shop for food close to home using a combination of objective and self-reported measures while simultaneously exploring disparities in geographic food access by race/ethnicity, urbanicity, and motivations of store selection among a lower-income, racially/ethnically diverse cohort in Central Texas. In this analysis, objective measures of the neighborhood food environment and self-reported shopping behavior data were utilized to examine if 1) members of a racially/ethnically diverse, predominantly lower-income cohort reported shopping at supermarkets closest to their home, 2) differences in overall distance traveled to self-reported supermarkets varied by race/ethnicity, urbanicity, and motivations of store selection, and 3) differences in excess distance beyond nearest grocery store varied by race/ethnicity, urbanicity status, and motivations of store selection.

Materials and methods

Study design and population

This cross-sectional analysis utilized baseline survey data from the FRESH-Austin study.¹⁸ The parent study has been described elsewhere, but to provide a brief summary: FRESH-Austin was a multi-level evaluation of a healthy food access intervention in Central Texas, and included a 3-year-long cohort study, assessments of the built and food retail environments, and an agent-based model.^18,51 Cohort participants (N = 400) were recruited and completed a baseline survey in October 2018–2019. The analytic sample for this study was restricted to residents of Travis County (seven individuals lived outside of Travis County and were excluded from the sample) and thus included 393 cohort participants who lived in Travis County, and met the parent study inclusion criteria of being 18 or older, the primary food shopper for their household, and spoke English or Spanish. Written informed consent was obtained from all participants in English or Spanish, and all participants were compensated for their time with an incentive. Additionally, data regarding food retail locations were provided from the City of Austin’s Food Environment Analysis (FEA).⁵² This study was approved by [Deidentified] Institutional Review Board for [Deidentified] (Deidentified IRB #).

Variables of interest

Geographic food access

Geographic food access was assessed through different measures. Using GIS data from the City of Austin’s FEA, all food retail locations in the county were geocoded and categorized by type of food retail. Home address and the address of the most frequently utilized supermarket were reported by participants via a survey and were geocoded.

Continuous variables measuring the street-network distance (in miles) between the participant’s home address location and the nearest supermarket, as well as to their most frequently used supermarket location were calculated. Next, the ratio of the distance to the nearest supermarket and the distance to the supermarket each participant reported using the most was calculated. Using this ratio, participants were then categorized as follows: 1) those who use their nearest supermarket, 2) those who use a supermarket that is somewhat further than their nearest store (1.1–5 times the distance), and 3) those who use a supermarket that is considerably further away from their nearest supermarket (over 5 times the distance). Geocoding and distance calculation tasks were run using ArcGIS version 10.7.1, Google Distance Matrix API, and R.

Racial/ethnic and urbanicity-related indicators

The main demographic indicators of interest for this analysis were race/ethnicity, and urbanicity. Race/ethnicity was self-reported by the participants.¹⁸ For descriptive purposes, race/ethnicity is a four-category, mutually exclusive variable including “Non-Hispanic White,” “Black,” “Hispanic,” and “Other.” This variable is then dichotomized as “Non-Hispanic White” and “People of Color” due to sample size issues. Urbanicity was determined based up on the zip code of the resident. A population density-based measure was determined and categorized based on the Census definition and the US Department of Defense definition of urban areas based on population density.^13,15,53 Zip codes with a population density of over 3,000 people per square mile were categorized as urban, zip codes with a population density between 1,000 and 3,000 people per square mile were categorized as peri-urban, and zip codes with a population density of less than 1,000 people per square mile were categorized as rural.^19,50,53

Primary motivation for store selection

In order to examine factors related to store selection choices, the FRESH baseline survey included a variety of questions related to motivating factors when choosing where to shop for groceries.¹⁸ The survey included adapted questions from the Eating and Health Module of the American Time Use Survey and asked participants to report the importance of the following factors when deciding where to shop: quality of store, quality of food, variety, cultural variety, price, convenient location, and availability of non-food items.⁵⁴ The measures had a 5-item response option, and binary variables were created to identify if the factor was the main factor for store selection based on whether participants reported that factor as very (#1) or somewhat (#2) important to them when deciding where to shop for food.

Covariates

Based on the literature, numerous covariates were included in the analysis, including income, utilization of food assistance programs in the last year, and main mode of transportation, which were all included in the FRESH Baseline survey.¹⁸ Income was reported as a categorical variable, as under $25,000, between $25,000–44,999, $45,000–$65,000, and over $65,000. Participants were asked if they utilized the Supplemental Nutrition Assistance Program (SNAP) and Women, Infants, and Children (WIC) program in the last year, respectively. Due to limited sample sizes, if a participant reported using either SNAP, WIC, or both programs in the last year, they were classified as yes in the binary variable. The main mode of transportation was reported as personal car, ride share car or taxi, bicycle, walking, and public transit bus; however, this was then dichotomized into personal car or other mode of transportation for regression analyses due to sample size.

Analysis plan

Descriptive statistics, mean differences, and unadjusted and adjusted multivariate linear regression were used in order to better understand the relationship between geographic food access and behaviors by race/ethnicity, urbanicity, motivations of store selection, and other covariates. Mean differences and linear regression were used to examine differences in continuous distance to closest and self-reported supermarket by race/ethnicity, urbanicity, motivations of store selection and other covariates, including income, utilization of food assistance in the last year, and main mode of transportation. Potential interactions between variables were assessed using Wald tests for interaction. Additionally, descriptive statistics were used to examine if there were differences in magnitude of distance traveled beyond their closest supermarket (categorical variable analysis) by race/ethnicity, urbanicity status, and main shopping motivation. Statistical analyses were performed utilizing Stata version 16, ArcGIS, and R.^55–57

Results

Description of the sample

The sample of the parent study (N = 400), described elsewhere, was predominantly Hispanic (54.4%), female (70.5%), and lower income (52.6% under $45,000 annual income).¹⁸ The sample had a similar composition in the slightly restricted sample used for this analysis (N = 393). Descriptive statistics for the variables of interest for the analysis are presented in Table 1. The sample was majority Hispanic (54.1%), almost a third of the sample was non-Hispanic White (32.31%), and just over 10% of the sample identified as Black. Over half of the participants resided in an urban zip code (55.4%), and earned less than $45,000 in annual income in 2017 (52.0%). The majority of participants used a personal car as their main mode of transportation (91.1%) and had not used SNAP and/or WIC in the last year (78.4%). In terms of shopping motivations, the majority of participants reported quality of items as their main reason for choosing where to shop for groceries (76.1%), followed by price (57.51%) and availability of non-food items (45.55%).

Table 1. Description of race/ethnicity, urbanicity, demographic characteristics, motivations of store selection, and geographic food access of the sample.

N=393	Variable N	%	Mean		SD
Race/Ethnicity
Non-Hispanic White			126	32.31
People of Color (Hispanic + Black + Other)			264	67.69
Hispanic			211	54.1
Black			40	10.26
Other			13	3.33
Urbanicity
Urban			216	55.38
Peri-Urban			92	23.59
Rural			82	21.03
Geographic Food Access (Continuous, in Miles)
Average Distance to Closest Supermarket					1.66	1.3
Average Distance to Self-Reported Utilized Supermarket					5.26	4.67
Geographic Food Access (Categorical)
Shopped at Closest Supermarket			73	18.6
Shopped at Supermarket Somewhat Far from Closest Supermarket (1.1-5x Distance)			208	52.9
Shopped at Supermarket Very Far from Closest Supermarket (>5x Distance)			112	28.5
Income
Under $25,000			86	22.93
$25,000-$44,999			109	29.07
$45,000-$65,000			69	18.40
$65,000+			111	29.60
Utilization of SNAP and/or WIC in Last Year
Yes			85	21.63
No			308	78.37
Main Mode of Transportation
Personal Car			358	91.09
All Other Modes of Transportation			35	8.91
Main Motivation of Store Selection
Quality of Items			299	76.08
Price			226	57.51
Variety of Items			134	34.10
Quality of Store			63	16.03
Cultural Variety			41	10.43
Availability of Non-Food Items			179	45.55
Convenient Location			291	74.05

Participants on average lived 1.66 miles away from the closest supermarket to their home but traveled on average 5.26 miles to their preferred supermarket. Less than 19% of the participants reported shopping at their closest supermarket. Just over half of participants shopped at a supermarket somewhat far from their closest supermarket (52.9%), and over a quarter shopped at a supermarket very far from their closest supermarket (28.5%).

Results from univariate and multivariate linear regression models

Mean differences and linear regression models were conducted to explore differences in distance traveled to the closest supermarket to home and to self-reported utilized supermarket by race/ethnicity, urbanicity status, motivations of store selection, and other covariates. There were no statistically significant differences in distance to closest supermarket by race/ethnicity or any other factor other than urbanicity by mean differences or in the adjusted multivariate linear regression models, respectively (rural participants lived farther away than urban participants [Mean differences β = 0.50, SE = 0.16, p < 0.01; Adjusted Linear Regression β = 0.48, SE = 0.18, p = 0.01]).

However, there were statistically significant differences in the mean differences and linear regression models examining distance traveled to self-reported grocery store. Findings of the mean difference analyses are presented in Table 2. Statistically significant mean differences were found by race/ethnicity, with Hispanics participants (β = 1.26, SE = 0.51, p = 0.02) and Black participants (β = 2.71, SE = 0.84, p < 0.01) had significantly higher distances to their self-reported utilized supermarket than Non-Hispanic White participants. Additionally, participants living in peri-urban zip codes (β = 1.77, SE = 0.37, p < 0.001) and rural zip codes (β = 9.29, SE = 0.38, p < 0.001) had significantly longer distances to their self-reported utilized supermarket than participants who were residents of urban zip codes. Also, participants who had used SNAP and/or WIC in the last year traveled further to their self-reported utilized supermarket than participants who had not used SNAP and/or WIC in the last year (β = 0.22, SE = 0.02, p < 0.001). However, there were no significant differences by income, main mode of transportation, or any of the motivating shopping factors related to grocery store selection.

Table 2. Mean differences findings exploring distances to self-reported supermarket and demographic and motivation of store-selection-related factors.

	β	SE	p
Race/Ethnicity (Referent = Non-Hispanic White)
Hispanic	1.26	0.51	0.01
Black	2.71	0.84	0.001
Urbanicity (Referent = Urban)
Peri-Urban	1.77	0.37	<0.001
Rural	9.29	0.38	<0.001
Income (Referent = Under $25,000)
$25,000–$45,000	−0.43	0.67	0.52
$45,000–$65,000	0.08	0.75	0.91
Over $65,000	0.02	0.67	0.98
Utilization of SNAP and/or WIC in Last Year (Referent = No)
Yes, used SNAP and/or WIC in Last Year	0.22	0.02	<0.001
Main Mode of Transportation (Referent = Car)
All Other Modes of Transport	−1.30	0.83	0.12
Binary Variables for Main Motivation of Store Selection
Quality of Food	−0.04	0.55	0.94
Price	−0.20	0.48	0.68
Convenient Location	−0.21	0.54	0.70
Variety	0.14	0.50	0.77
Cultural Variety	0.15	0.77	0.85
Quality of Store	0.87	0.65	0.18
Availability of Non-Food Items	0.83	0.47	0.08

Potential interactions between variables were also examined using Wald tests for interaction. There were significant interactions between race/ethnicity and numerous factors related to grocery store selection, including quality of food (p = 0.02), convenient location (p = 0.01), availability of non-food items (p = 0.01), variety (p = 0.02), cultural variety (p = 0.02), quality of store (p = 0.01), and price (p = 0.01). Due to these interactions, multivariate linear regression analyses were stratified by race/ethnicity and presented in Table 3. Among non-Hispanic white participants, urbanicity and convenient location of store played significant roles in distance traveled to self-reported utilized supermarket. Non-Hispanic white participants living in rural zip codes (β = 8.97, SE = 0.74, p < 0.001) had significantly longer distances to their self-reported utilized supermarket than participants who were residents of urban zip codes. Also, non-Hispanic white participants who reported convenient location as an important factor related to grocery store selection did not travel as far to their self-reported utilized supermarket than those who did not report convenient location as an important factor (β= −1.46, SE = 0.48, p < 0.01).

Table 3. Adjusted linear regression analytic findings exploring distances to closest supermarket and demographic and motivation of store-selection-related factors stratified by race/ethnicity.

	Non-Hispanic White Participants				Hispanic, Black, and/or Other Participants
	β	SE		p	β	SE		p
Urbanicity (Referent = Urban)
Peri-Urban	0.95	0.67	0.16		2.26		0.48		<0.001
Rural	8.97	0.74	<0.001		9.46		0.50		<0.001
Income (Referent = Under $25,000)
$25,000–$45,000	−1.53	1.06	0.15		−0.87		0.51		0.09
$45,000–$65,000	−1.74	1.03	0.10		−0.17		0.67		0.80
Over $65,000	−1.21	1.01	0.23		1.45		0.64		0.03
Utilization of SNAP and/or WIC in Last Year (Referent = No)
Yes, used SNAP and/or WIC in Last Year	−1.56	1.10	0.16		0.67		0.72		0.35
Main Mode of Transportation (Referent = Car)
All Other Modes of Transport	−0.09	0.86	0.92		0.21		0.48		0.67
Binary Variables for Main Motivation of Store Selection
Quality of Food	−0.41	0.71	0.57		0.31		0.80		0.70
Price	−0.10	0.64	0.88		0.62		0.83		0.45
Convenient Location	−1.46	0.48	<0.01		−0.53		0.51		0.30
Variety	0.61	0.88	0.38		0.16		0.83		0.84
Cultural Variety	−0.65	1.04	0.53		0.31		0.96		0.75
Quality of Store	0.79	0.80	0.33		0.43		0.90		0.63
Availability of Non-Food Items	0.33	0.48	0.50		0.13		0.42		0.75

Among participants who were Hispanic, black and/or other races and ethnicities, urbanicity and income played significant roles in distance traveled to self-reported utilized supermarket. Among participants who identified as people of color living in peri-urban zip codes (β = 2.26, SE = 0.48, p < 0.001) and rural zip codes (β = 9.46, SE = 0.50, p < 0.001) had significantly longer distances to their self-reported utilized supermarket than participants who were residents of urban zip codes. Also, among this group, participants who had an annual income over $65,000 in 2017 traveled further (β = 1.45, SE = 0.64, p = 0.03) than those who earned less than $25,000 in 2017. However, there were no significant differences in travel by shopping motivations among Hispanic, Black, and/or other races/ethnicities in this sample.

Results from descriptive statistics of categorical measures of excess distance traveled

Descriptive statistics were also conducted to explore differences in the excess distance beyond the closest supermarket traveled to the self-reported utilized supermarket by race/ethnicity (as shown in Fig. 1), urbanicity (Fig. 2), and main shopping motivation (Fig. 3). Although tests assessing statistically significant differences could not be completed due to insufficient cell sample size, there are notable differences by race/ethnicity, urbanicity status, and main shopping motivation. More specifically, 12.50% of the respondents of color in the sample traveled over 5 times the distance to their closest supermarket to their preferred grocery store, compared to 7.14% of the non-Hispanic white participants. For urbanicity differences, participants residing in urban zip codes had the highest proportion of shopping at their closest supermarket (>25%), while rural participants had the highest proportion of shopping at a supermarket over 5 times the distance to their closest supermarket (30.49%). With regard to shopping motivations, the majority of respondents across all categories shopped somewhat farther than their closest supermarket (1.1–5 times the distance), but respondents who reported that cultural variety was their main shopping motivation had the highest proportion (>80%). Additionally, respondents who stated that the availability of non-food items was one of their main shopping motivations had the highest proportion of those who shopped at their closest grocery store (22.35%).

Figure 1. Distribution of excess distance traveled to utilized supermarket by race/ethnicity.

Figure 2. Distribution oF excess distance traveled to utilized supermarket by urbanicity.

Figure 3. Distribution of excess distance traveled to utilized supermarket by main shopping motivation.

Discussion

Summary of findings

Our study, which utilized a racially/ethnically diverse, lower-income sample from a preexisting cohort in Central Texas, contributes valuable evidence that adds to the growing body of evidence that individuals do not necessarily shop for food at the supermarket closest to home. Specifically, our study found that the majority of participants in our cohort do not shop at the supermarket closest to home. Additionally, this study presents evidence of the role of race/ethnicity, urbanicity, and shopping motivations in geographic food access, in terms of distance traveled to self-reported utilized supermarket (linear regression findings), as well as the excess distance traveled beyond the closest supermarket to home (descriptive statistics).

This study presents findings that are consistent with the urban-specific literature conducted in cities such as Philadelphia, Los Angeles, Atlanta, and others that found that lower-income residents rarely shopped at the closest food retail location.^{33,36,44–46} This phenomenon is also corroborated by research conducted in rural settings that found that individuals may bypass their closest supermarket/grocery store for other motivations.^13,15,16,47 Thus, this work adds to the body of literature highlighting the need to include objective measures of the food environment paired with community-specific data in order to best understand the nuances of complex issues such as food access and shopping behaviors.^26,33,35,58 Additionally, this was found among a Central Texan sample that was racially/ethnically diverse, lower income (but did have higher-income participants for variation), and had residents that lived across the urban-rural continuum, presenting helpful region-specific findings.

Furthermore, this analysis highlights the roles that race/ethnicity, urbanicity, and shopping motivations play when assessing geographic access to food. While there were significant differences by urbanicity for distance to the closest supermarket, with rural participants traveling further than urban participants, the magnitude of this finding was much larger for where participants actually reported shopping and varied by race/ethnicity. Specifically, with peri-urban and rural participants traveling significantly further than urban participants among participants of color, while only rural participants traveled significantly further than urban participants among non-Hispanic white participants. These racial ethnic differences are helpful insights to literature that found racial/ethnic differences among solely urban and/or rural samples.^16,43–45

Additionally, there were significant interactions between race/ethnicity and shopping motivation-related factors. The directionality of the association of distance to utilized supermarket varied by race/ethnicity for quality of food, price, and cultural variety. Also, the magnitude of the association of distance to utilized supermarket varied by race/ethnicity for convenience, variety, quality of store, and availability of nonfood items. These are helpful insights to the growing body of literature that has found that shopping motivations can vary by race/ethnicity and other sociodemographic factors such as income.^39–45 Our findings, and the larger literature, suggest that race/ethnicity needs to be considered when exploring factors related to access to food such as affordability, accessibility, and other components. These findings echo the call of scholars such as Ver Ploeg and Wilde (2018) and Singleton and colleagues (2020), who have discussed the role of household-level factors and the importance of exploring intersectionality when examining geographic access to food, respectively.^4,37 At the time of the development of this manuscript, this is the first study to assess all of these factors simultaneously with a lower-income, racially/ethnically diverse sample with residents across the urban, peri-urban, and rural settings.

Additionally, this study echoes the critiques brought forth by Chen and Kwan (2015) that discuss the role of the uncertain geographic context problem in food access research (UGCoP).³¹ As articulated previously, the authors critiqued that the current literature base assumes that an individual’s immediate food environment is centered around proximity to their home address, and this assumption could be erroneous, insufficient, and could be limiting the scope of our understanding of food access.³¹ Our findings add to a field of research that call for greater exploration of the food environment outside of just the immediate neighborhood in a diversity of settings.^{16,32,37,38,40,41,44,45,59,60}

Future food access research should potentially adopt a model found in the physical activity literature that discusses the role of “Activity Spaces.”^61,62 In this literature, Perchoux and colleagues (2013) state that this paradigm shifts the focus of geographic analysis from the neighborhood-level to a more comprehensive approach informed by wearable devices to account for multiple locations that an individual visits over time.⁶¹ This approach would enable the food access literature to look past solely looking at the role of geographic food access, and explore other aspects of the access definition such as affordability, cultural acceptability, availability, and other domains.^2,63 Additionally, this methodological assessment could assist in the gap in the literature highlighted by Wilde and colleagues (2021) and others regarding the need for greater examination of shopping behavior data is needed among in order to determine if preferred shopping locations are driven by affordability, availability of culturally relevant products, availability of other products and services, other factors pertaining to access or a combination of the above factors.^{38,40,41,43–45,59,60,64,65}

Strengths and Limitations

Strengths of Study

There are several strengths to this study. Previous studies typically have only utilized census data with food retail, and not additional data sources, and researchers have often stated the need for utilizing multiple source data such as geographic and survey data, as was done with this study.^17,26,35,66 This work also provides a valuable contribution to the literature by examining various household-level factors such as race/ethnicity, urbanicity (including urban, peri-urban, and rural), motivations of store selection, income, and other covariates simultaneously in order to account for potential intersectionality, as other scholars have highlighted the need for in the literature.^4,37,38 Additionally, this study was conducted in the Central Texas region, which enables a setting that has variation in race/ethnicity, urbanicity, income, and other factors in one geographic region. This is unique to other studies that have explored similar questions in solely urban, or rural contexts.¹⁶ These findings provide new and unique insight into how individuals engage in their food environment and make a strong case that further research is needed.

Limitations of study

However, there are limitations to this analysis. This study employed a cross-sectional study design; therefore, causality cannot be inferred. Additionally, this study utilized self-reported measures for utilized shopping location, which could add issues with recall bias. Future work should include self-reported and objective measures of utilized shopping locations, such as with a GPS device. Additionally, while recruitment of participants occurred in specifically selected communities due to their demographic factors and community resources (or lack thereof), participants had to agree to participate in the study, therefore it may not be the most accurate depiction of shopping behaviors of all residents in these communities.

Thus, the resulting sample predominantly residing in the Eastern Crescent of Austin (which is mainly low-income) were majority Hispanic; therefore, these findings may not be generalizable to other communities or other areas in the Central Texas region.⁶⁷ Thus, similar work is needed employing a more representative sample that also simultaneously employs an intersectional approach and utilizes shopping behavior and motivation data. Yet, this regional corroboration with individual-level data is needed in the literature, and even with the identified limitations, this study makes valuable and needed contributions to the literature.

Public Health Implications and Future Research

Our findings contribute evidence of racial/ethnic, and urbanicity disparities in geographic food access and suggest that strategies that focus solely on improving geographic food access may be overlooking other determinants of food shopping behavior, such as food affordability, quality, and cultural adequacy. Therefore, this work can help future researchers and planners consider how to better address geographic food access issues and demographic considerations. Given the cross-sectional nature of this study, future research examining geographic food access, shopping behaviors, and demographic characteristics is needed. Incorporating survey with GPS-monitoring of habitual spatio-temporal activity by individuals could help advance the field of geographic food access research. This could allow the examination of the impact of an individual’s working location, commuting route, or other non-home-based activities on the selection of locations where they shop for food.^14,34,58,68 Ideally, future work examining geographic food access should include objective and community-specific measures in order to examine these complex issues with greater nuance.

Conclusions

In conclusion, our study adds to the geographic food access literature in several ways, including contributing evidence to the growing body of literature challenging the assumption that individuals shop for groceries close to home in a racially/ethnically diverse region across urban, peri-urban, and rural contexts. Our work also demonstrates the importance of using objective and community-specific measures when exploring geographic food access, and presenting evidence of racial/ethnic, urbanicity-, and shopping motivation-associated differences in geographic food access. Although some researchers and policymakers focus on strategies to improve geographic food access, other domains of access (affordability, quality, cultural adequacy) need greater consideration.

Ethics Approval and Consent to Participate

The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board (or Ethics Committee) of the UTHealth School of Public Health Institutional Review Board (IRB HSC-SPH-18-0233; overarching study approved − 5/19/18. Informed consent was obtained from all subjects involved in the study and was offered signed on forms presented in English and in Spanish. Blank consent forms can be shared by request and contacting the authors.

Consent for Publication

As previously stated, the study and possible publication plans were conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board (or Ethics Committee) of the UTHealth School of Public Health Institutional Review Board (IRB HSC-SPH-18-0233; overarching study approved − 5/19/18, COVID-19 specific protocol approved − 4/23/20). Informed consent was obtained from all subjects involved in the study and consent included consent to publish de-identified aggregated findings and was offered signed on forms presented in English and in Spanish.

Availability of Data and Materials

All data and instruments utilized in this study can be shared by request and contacting the authors.

Author Contributions

Conceptualization, K.J.; methodology, K.J., D.S., A.vdB., N.R.; formal analysis, K.J.; writing – original draft preparation, K.J.; writing – review and editing, D.S., N.R., A.vdB, D.M.H, A.N.; supervision, A.vdB., N.R., D.S., D.M.H.; project administration, A.N.; funding acquisition, A.vdB., D.S., N.R., J.C., P.L. All authors have read and agreed to the published version of the manuscript.

Acknowledgments

We would like to acknowledge members of the community, without which this research would not have been possible. We would also like to acknowledge various members of the research team who assisted with the data collection including Martha Diaz, Nika Akhavan, Christine, Jovanovic, Shelby Flores-Thorpe, Gozie Ibeji, and others.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research was funded by the Foundation for Food and Agriculture Research, 560815-42731 (Sustainable Food Center, Subcontract – UTHealth School of Public Health). Preparation of this manuscript by KJ was funded in part by The National Cancer Institute (NCI)/National Institutes of Health (NIH) Grant—NCI/NIH Grant T32/CA057712, awarded to the University of Texas Health Science Center at Houston School of Public Health Cancer Education and Career Development Program.