Race, Socioeconomic Status, and Food Access in Two Predominantly White Cities: The Case of Lansing, East Lansing, and Surrounding Townships in Michigan

Abstract

Access to fresh, healthy, affordable foods is a pressing concern in cities worldwide. American cities are no exception. Although many scholars study food access in large cities, small and mid-sized American cities can provide valuable information about inequities in the food system. This paper focuses on two adjoining, racially mixed Mid-Michigan cities—Lansing and East Lansing. It examines the extent to which different food outlets exist in the cities and surrounding townships. It probes the following questions: (1) How are food outlets distributed throughout the cities and suburbs? (2) What is the relationship between neighborhood demographic characteristics and the distribution of food outlet types? We collected data on food outlets from September 2020 to June 2022 using Data Axle as our primary source of information. We used ArcGIS 10.8.1 for the spatial mapping and SPSS 28 for statistical analyses. We conducted regression analyses to identify the difference in the likelihood of finding food retailers in census tracts where 0–20% of the residents were People of Color (VL-POC), 20.01–40% of the inhabitants were People of Color (L-POC), 40.1–60% of the residents were People of Color (H-POC), and more than 60% of residents were People of Color (VH-POC). There were 1647 food outlets in the study area: 579 were in Lansing, 220 were in East Lansing, and the remaining 848 were in the surrounding townships. Restaurants dominated the food landscape, while small groceries and convenience stores were the grocery sector’s most common food outlet types. Supermarkets and large grocery stores comprised only 5.6% of the study area’s food outlets. The study finds a nonlinear relationship between the racial composition of census tracts and the prevalence of food outlets. The VH-POC census tracts had very few food outlets. For instance, the tracts had no supermarkets, mass merchandisers or supercenters, small grocery or convenience stores, pharmacies or drug stores, or farmers’ markets. The findings illustrate the diversity and complexity of the Lansing–East Lansing metropolitan area’s food landscape.

1. Introduction

Food insecurity and food access are global concerns. Although global food security improved from 2012 to 2019, since peaking in 2019, it has declined, and the result has been rapidly rising food prices, widespread hunger, and undernourishment [1]. According to the Global Hunger Index, although worldwide hunger fell rapidly from 2000 to 2014, the rate of decline has slowed noticeably since then [2]. The coronavirus disease (COVID-19) pandemic disrupted food production and distribution systems and contributed to declining food security. Consequently, an estimated 828 million people worldwide were undernourished in 2021 [2,3].

Food insecurity is prevalent in the U.S., and international comparisons point to some troubling statistics. The Global Food Security Index measures four aspects of food security and scores each on a scale of 0–100: affordability, availability, quality and safety, and sustainability and adaptation. Each country is given a composite score, as well as a score for each of the factors. In 2022, the U.S. ranked 13th among the 113 countries studied on the index, with an overall score of 78.0. The U.S. scored 87.1 for food affordability, 65.1 for availability, 88.8 for food quality and safety, and 69.4 for sustainability and adaptation [1]. Therefore, it is unsurprising that food insecurity—especially food access—is a growing concern in rural areas, small towns, and cities across the U.S.

A robust body of research addresses these issues. Researchers pay much attention to urban food insecurity and food access. However, investigators prefer to study large or famous cities, hyper-segregated cities with well-defined ethnic enclaves and neighborhoods, and ones with stark socio-economic contrasts and large populations of People of Color (POC). This is a common feature of food access studies in Michigan. Consequently, several of the state’s food access studies are conducted in places like Detroit and Flint.

For instance, Detroit and Flint have experienced substantial population declines since hitting their peak in the 1950s and 1960s. Between 1950 and 2021, Detroit lost 65.8% of its population, while Flint lost 59.1% between 1960 and 2021. In comparison, Lansing lost 14.3% of its population since reaching its zenith in 1970; East Lansing lost 8.8% since peaking in 1980 [4,5,6,7,8,9,10,11,12,13,14,15,16].

East Lansing has the highest poverty level of the four cities and Lansing the lowest (37.9% and 22.7%, respectively), while Detroit’s poverty rate is 31.8%, and Flint’s rate is 35.5%. However, the median household incomes of Lansing and East Lansing are noticeably higher than Detroit’s and Flint’s. So, the median household income is USD 44,233 in Lansing and USD 40,800 in East Lansing. It is USD 34,762 in Detroit and USD 32,358 in Flint [4,5,6,7,8,9,10,11,12,13,14,15,16].

East Lansing and Lansing are predominantly White cities, while Flint and Detroit are predominantly Black. Hence, East Lansing is 69.8% White, and Lansing is 51.4% White. On the other hand, Flint is 56.7% Black, and Detroit is 77.9% Black [4,5,6,7,8,9,10,11,12,13,14,15,16].

East Lansing and Lansing have higher levels of educational attainment than Flint and Detroit. Roughly 98% of East Lansing’s population and 90.5% of Lansing’s have completed high school. In Flint, 84.8% of the population and 82.6% of Detroit’s residents graduated from high school. East Lansing also has the highest rate of college completion. That is, 68.3% of East Lansing’s residents have completed college; 28.7% of those in Lansing have completed a bachelor’s degree. Although Detroit and Flint have major universities, 16.2% of Detroit’s residents have completed a bachelor’s degree; 12.1% of Flint’s residents have done likewise [4,5,6,7,8,9,10,11,12,13,14,15,16].

Our study breaks with convention and examines lesser-known urban areas. It focuses on two small to mid-sized Michigan cities and the townships around them. The study area is not characterized by the extremes common in urban food access studies. This paper reports the findings of a study of food access in the Lansing–East Lansing metropolitan area of Michigan. It examines the extent to which different types of food outlets are present in the cities and adjacent suburbs and probes the following questions: (1) How are food outlets distributed throughout the study cities and suburbs? (2) What is the relationship between neighborhood demographic characteristics and the distribution of food outlet types?

There are several dimensions to food access, including physical access, financial access, the quality of food, the nutritious content of food, cultural access, familiarity with food, the safety of food, and language barriers [17]. As mentioned earlier, most of the food access studies conducted in Michigan focus on Detroit [17,18,19,20,21,22,23,24,25,26,27,28], and Flint [29,30,31,32,33,34,35,36,37,38,39,40]. Detroit, Michigan’s largest city, is known as an automotive manufacturing center. It is also known for its music production and fiscal challenges [17]. Although Flint is smaller than Lansing, it is also known for manufacturing automobiles and lead-contaminated water [41].

Despite being the state capital, other cities often overshadow Lansing and its smaller neighbor, East Lansing. Nevertheless, a compelling case can be made for conducting a food access study in the locale. Lansing and East Lansing have not hemorrhaged jobs and people to the same extent as Detroit or Flint. Hence, Lansing–East Lansing provides an opportunity to examine food access in an area where residents with a high likelihood of food insecurity are present in the study area, but do not dominate the population.

Studies report that Blacks and Hispanics/Latinx are more prone to food insecurity than other racial and ethnic groups [42,43,44]. Though Lansing and East Lansing are predominantly White cities, they have neighborhoods where Blacks and Latinx inhabitants comprise the majority of the residents.

East Lansing is home to a major university, but it should not be overlooked as a place to examine food access. Many of the students live off campus, and food insecurity is high among college students. For instance, a national study of nearly 86,000 college students found that 45% experienced food insecurity [45]. Later discussion shows that many Michigan State University students say they are food insecure [46].

Moreover, both Lansing and East Lansing have histories of racially exclusionary housing policies and practices that mirror other midwestern cities [47]. Those historic practices left a legacy that still influences who lives where in the cities.

Consequently, Lansing and East Lansing can provide insights into the question—Do People of Color who live in majority-White cities have similar access to food outlets as their White counterparts? As mentioned earlier, many studies examine racial differences in food access in predominantly Black or Latinx cities. Lansing and East Lansing are neither. Therefore, we are curious to find out how race is related to food access. We also want to know how our findings compare with earlier studies.

1.1. Historical Context

Lansing and East Lansing are on the ancestral land of the Anishinaabeg—Three Fires Confederacy of the Ojibwe, Odawa, and Potawatomi people [48,49]. The Ojibwe are also known as the Chippewa, Ojibway, or Ojibwa. Members of the Fox tribe also lived in East Lansing [50].

The first European explorers saw the Mid-Michigan region in 1790. After decades of fighting and appropriating Indigenous land, the United States government forced the Chippewa to cede the land Lansing and East Lansing are now on in the 1819 Treaty of Saginaw [49,51,52]. The densely forested area was surveyed in 1827 and put up for sale three years later. Two brothers from Lansing, New York, acquired land at the intersection of the Red Cedar and Grand Rivers in 1835 and named the area Biddle City. Upon their return to Lansing, New York, the brothers sold land plots by deceiving purchasers and claiming that Biddle City was already developed [53]. Sixteen men bought plots of land in Biddle City. Upon discovering that the land was undeveloped, they decided to stay; these settlers renamed the area Lansing Township [49,53,54,55].

Lansing, Michigan’s capital, is the seventh-largest city in the state. Lansing became the state capital in 1847 when the State Legislature voted to move the capital from Detroit out of fear of a British invasion via Canada [14,49,51,53]. During the late 1840s and 1850s, the city’s residents opposed slavery and became a stop on the Underground Railroad for those seeking freedom [56]. Lansing became an industrial city in 1897 with the founding of the Olds Motor Vehicle Company. Other automotive companies like the Reo Motor Car Company operated in Lansing from 1904 to 1975, and the short-lived Clarkmobile from 1902 to1904 [53,57,58,59]. In 1956, Lansing was 15 square miles (39 km²); it was 39.14 square miles (101.3 km²) in 2020 [8,53]. Today, government, manufacturing, insurance, health care, education, and banking drive the city’s economy [53].

East Lansing, the 27th-largest city in the state, is home to Michigan State University. The university started as a scientific and agricultural college in 1855 and was established in East Lansing in 1857 [14,49,60]. It became a land-grant college in 1862 [61]. To provide more housing and create a livelier off-campus atmosphere for the still rural campus, professors at the university created and planned Collegeville. Later, the university supported the development of College Delta. These communities eventually merged to form the city of East Lansing in 1907 [49,60,62].

1.2. The Demographic Characteristics of the Study Cities

Although Kaminski [63] argues that the Greater Lansing Metropolitan Area is one of the fastest-growing regions of the Midwest, there are signs that population growth might be slowing. The Greater Lansing Metropolitan Area, consisting of Ingham, Eaton, Clinton, and Shiawassee counties, had a population of 502,444 in 1990 and 541,297 in 2020. However, the population dipped to 540,281 in 2021 [6,64].

The two core cities—Lansing and East Lansing—have experienced population declines. Lansing’s population grew steadily from 1860 to 1900; it grew dramatically over the next seven decades, going from roughly 16,500 to its peak of 131,403 in 1970 [14,16]. The city’s population fell to 114,297 in 2010 and has fluctuated since then. The population estimates showed a temporary uptick between 2018 and 2019. During this time, the population increased from about 116,699 to 117,159. However, the population dropped to 112,644 in 2020. The 2021 American Community Survey (ACS) indicates the city’s population is 112,684 [8,14].

The population of East Lansing grew rapidly from 1940 to 1980, when it peaked at 51,392. By 2010, the population had dropped to 48,579. The city’s population declined slowly; it was 48,374 in 2018 and increased slightly to 48,729 in 2019. The 2020 census reports that the city’s population is 47,741. The ACS estimates that East Lansing’s population was 46,854 in 2021 [9,14].

Table 1. Population Characteristics of Lansing and East Lansing, Michigan.

Population Characteristics	Michigan		Lansing		East Lansing
	Population	Percent	Population	Percent	Population	Percent
Total Population	10,077,331	100.00	112,644	100.00	47,741	100.00
White alone (not Latinx or Hispanic)	7,295,651	72.40	57,838	51.35	33,331	69.82
Black alone (not Latinx or Hispanic)	1,358,458	13.48	25,376	22.53	5732	12.01
Hispanic or Latinx	564,422	5.60	15,467	13.73	2477	5.19
Native American or Alaska Native	47,406	0.47	504	0.45	123	0.26
Asian	332,288	3.30	4732	4.20	4202	8.80
Pacific Islander	2603	0.03	32	0.03	49	0.10
Other	37,183	0.37	767	0.68	176	0.37
Two or more races	439,320	4.36	7928	7.04	1651	3.46

Compiled from: [7,8,9].

shows the racial characteristics of Michigan, Lansing, and East Lansing. Michigan’s population topped 10 million in 2020. Most of the state’s residents (72.4%) are White. Lansing is a racially/ethnically diverse city, wherein 51.4% of the population is White, 22.5% is Black, 13.7% is Hispanic or Latinx, and 4.2% is Asian. In contrast, 69.8% of East Lansing’s population is White, 12% is Black, 5.2% is Hispanic or Latinx, and 8.8% is Asian.

Thirteen Blacks lived in Lansing in 1850, and Blacks have been residents of the city since then [65]. Early Black settlers were forced to live in a small enclave around the Oldsmobile/General Motors factory, south of the city center and north of the Grand River [66]. Lansing became a hyper-segregated city, as evident in a 1969 housing discrimination lawsuit that decried yesteryears, when the city embraced segregationist policies and practices [67]. The lawsuit describes the extreme segregation of Blacks and Mexican Americans in Lansing. When the suit was filed, these two groups comprised about 11% of the city’s population, yet 65% of them resided in one neighborhood, where 75% of the housing was substandard or dilapidated [67]. The conditions described in the suit arose from the redlining of Black neighborhoods, denial of federally backed mortgages, the use of racially restrictive covenants to bar the rental or sale of property in White neighborhoods to Blacks, residential segregation, urban renewal projects, and the construction of the Interstate 496 highway through the heart of the Black community [47,66,68,69,70]. For instance, Felber [68] describes the use of racially restrictive covenants in Lansing to prevent Malcolm X’s family from taking ownership of property they purchased in 1929 in a White neighborhood.

Taylor [47], Aaronson et al. [71,72], and Sadler et al. [73] contain discussions of redlining, residential segregation, and housing discrimination. Between 1935 and 1940, the federal government-backed Home Owners’ Loan Corporation (HOLC) mapped cities and placed red lines around Black communities and green lines around White, Western European non-immigrant communities. Residents of redlined communities found it difficult to obtain loans or mortgages; it was also very hard to move from such neighborhoods. In contrast, residents of greenlined communities received loans to purchase, build, or repair their property. HOLC based its maps on the discriminatory practices and guidelines of the Federal Housing Authority (FHA). To view a copy of the 1934 HOLC redlining map of Lansing and East Lansing, see Carpenter [74]. Remnants of the segregation that was commonplace in the city can be found in the Towar Gardens Wall, which was once used to separate a Black neighborhood from a White neighborhood physically. The wall still stands [75].

Today, Lansing celebrates its racial diversity and is touted as one of Michigan’s most racially integrated cities [76]. The Dissimilarity Index (D) measures the extent to which a particular group is evenly distributed across census tracts in an urban area as a different group. The higher the D value between any two groups, the greater the likelihood that the two groups of residents live in different census tracts. The D score ranges from 0 to 100. The higher the score, the more segregated a community is. A score of 60–100 is considered very segregated, 31–59 is moderately segregated, and 0–30 reflects low levels of segregation [77]. An analysis of segregation in the city shows that between 1980 and 2020, Black–White, Hispanic–White, Black–Hispanic, Black–Asian, and Hispanic–Asian segregation decreased. Asian–White segregation increased from 1980 to 1990, but has declined steadily since then. The city currently has a Black–White D score of 26; the index is 18 for Hispanic–White, 32.3 for Asian–White, 16.4 for Black–Hispanic, 24.4 for Black–Asian, and 31.2 for Hispanic–Asian [77].

East Lansing had its share of residential segregation, redlining, and prohibitions on who could purchase homes in the city. Until the late 1960s, Blacks were barred from purchasing homes in the city [66,78,79]. In 2018, the city issued an apology for the racism Blacks faced when they tried to purchase homes prior to the passage of the 1968 Fair Housing Act [80]. Realtors and restrictive covenants embedded in property deeds prevented Blacks from buying homes in the city [78,80].

The state of racial integration in East Lansing differs from that of Lansing. An analysis of East Lansing’s dissimilarity indices shows that Black–White segregation rose between 1980 and 2000 but declined afterward. The D score was 29.3 in 2020. The Hispanic–White D score has fallen sharply over the four decades to 11.2 in 2020. White–Asian segregation increased significantly between 1980 and 1990 but has fallen steadily since then. The White–Asian D score is 29.7. The Black–Hispanic D score fluctuates, but, generally speaking, has risen; it is 29.3. The Black–Asian D score has risen dramatically to 39.7, while the Hispanic–Asian D score is 23 [77].

Despite their juxtaposition, East Lansing is a poorer city than Lansing. Lansing has a poverty rate of 22.7% and a median household income of USD 44,233. East Lansing has a poverty rate of 37.9% and a median household income of USD 40,800 [8,9].

2. Literature Review

2.1. Conceptualizing Food Insecurity

Food insecurity is a growing concern in the United States because of rising poverty, living expenses, unemployment, and the impacts of the COVID-19 pandemic. Many people lost jobs and income during the pandemic when workplaces like restaurants and stores closed. Loss of jobs and income also occurred when people became ill from the virus. The U.S. Department of Agriculture (USDA) defines food security as the ability of all members of a household to always have access to enough nutritious, safe, and affordable food at all times for health and vigor, without relying on emergency food assistance, scavenging, theft, or using other coping strategies. On the flip side, food insecurity occurs when there is limited or uncertain ability to access enough nutritious, affordable, and safe food to fulfill rudimentary nutritional needs [81,82,83].

Although the U.S. food insecurity rate had declined steadily for over twenty years, the pandemic reversed the trend [44,84,85]. Feeding America [86] reports that one in eight adults and one in seven children are food insecure in Michigan. In 2019, this meant about 1.3 million people in Michigan, including 305,000 children, were food insecure. The following year, the number of food insecure increased to 1.9 million people, including over 550,000 children [87]. The USDA assessed hunger in Michigan in 2020 and found that 11.8% of the households surveyed were food insecure [88]. The United Health Foundation [89] estimated that 12.9% of Michigan’s households were food insecure. Lansing and East Lansing are partly in Ingham County. The Greater Lansing Food Bank reported that food insecurity in Ingham County was 13.8% before the COVID-19 pandemic and was estimated at 19% in 2021 [90].

This article will analyze the distribution of food outlets in Lansing, East Lansing, and the surrounding townships. This research will identify the various food outlets in the study area and examine how different demographic factors are related to food outlet distribution.

2.2. Theorizing Neighborhood Food Environments

Many terms are used in food access research to classify different food landscapes. One of the most common terms, food desert, is often used to describe areas lacking access to supermarkets and large grocery stores that provide affordable, nutritious, and fresh foods [18,20,25,26,27,91,92,93,94,95,96,97]. Taylor & Ard [17] discuss the food desert research and critique the food desert and food swamp concepts. Despite growing criticism, the USDA uses the term food desert to identify areas with high food insecurity. Hence, low-income urban census tracts with no or few supermarkets, supercenters, or large grocery stores are sometimes classified as food deserts. A census tract is described as a food desert tract if its poverty rate is 20% or more or it has a median income that is less than 80% of the area’s median family income. To qualify, the tract must have at least 500 residents, and at least a third must reside more than a mile from a supermarket or large grocery store (10 miles in rural areas) [98,99,100,101,102,103].

Researchers describe a food swamp as a low-income area with an overabundance of unhealthy food retailers. The area usually has many mini marts, convenience stores, fast-food restaurants, liquor stores, and gas stations that sell food [104,105,106,107,108,109,110,111]. Scholars critiquing the food desert and food swamp theses find what they describe as food oases in low-income Communities of Color. A food oasis is an area with supermarkets, full-line grocery stores, and other food outlets offering affordable, culturally desired foods [112]. Taylor & Ard [17] and Taylor et al. [40] have detailed discussions of additional concepts, such as food grasslands, supermarket redlining, and food apartheid.

Despite the heavy focus on food deserts and the U.S. Department of Agriculture’s Food Access Research Atlas’ [113] identification of low-access census tracts, prior studies have not identified what we call extremely low-access (ELA) census tracts. These are tracts that contain none of the food outlets we studied.

We adopt a food systems approach to studying food access, informed by food justice and food sovereignty [17]. We conceptualize the cities we study as food systems and examine a variety of places where people grow, purchase, or obtain food. We take this approach because we believe that using proximity to supermarkets and full-line grocery stores as the sole or primary indicator of food access or food insecurity results in a misunderstanding of local food landscapes and understates food availability [17,91,114,115]. However, it should be noted that although this study examines an extensive array of food outlets, it does not study how food can be obtained through subsistence activities, like hunting, fishing, gathering, foraging, gleaning, scavenging, and similar means.

Having a just and fair food system facilitates food sovereignty. Jones et al. [116] identify five tenets of food sovereignty: (1) using agroecological principles in food production, (2) localizing food production and consumption, (3) promoting and practicing social justice and equity, (4) using traditional knowledge, and (5) transforming political and economic structures to further self-determination. Hence, food sovereignty means communities and individuals reduce or eliminate their dependence on industrial food systems and gain control over the means of food production, distribution, and consumption [17,24,114,116,117]. Some argue that food sovereignty is crucial to the survival of Communities of Color, particularly Blacks and Indigenous peoples [118]. Ergo, our study includes community gardens and urban farms—spaces where food sovereignty is often practiced.

Residents in communities with few supermarkets and full-line grocery stores obtain food from other venues. Other food outlet types may include takeout establishments, full-service restaurants, fast-food restaurants, mini markets, pharmacies, convenience stores, dollar stores, farmers’ markets, produce vendors, mobile food vans, community gardens, limited-assortment food stores, gas stations with food, bakeries, caterers, food banks, liquor stores, coffee shops, bars, banquet halls and hotels, ice cream parlors, meat markets, wholesalers, food pantries, and soup kitchens [17,40].

The pandemic impacted food landscapes by precipitating food outlet closures. Studies that probe pandemic food store closures have identified a vanishing food infrastructure characterized by a hollowing out of food outlets because food venues cease operations [29]. The researchers coined the terms to describe the extensive and, at times, concentrated food store closures they found in Flint and the surrounding townships. There were 173 closed food outlets; 81 were in Flint and 92 in the suburbs. There are signs of extensive closure of food outlets elsewhere. Other scholars, such as Yi et al. [119], who studied six New York City neighborhoods, found 2720 closed food outlets—most were in Chinese neighborhoods. Lowery et al. [120] studied a low-income, food-insecure neighborhood in San Diego, California, and identified 184 food stores; 23 closed during the pandemic. Bell & Taylor [29] use the term emergent food infrastructure to describe the opening of new food venues to replace closed facilities or as additional food outlets.

2.3. Measuring Food Access in Lansing and East Lansing

As mentioned before, there is limited research on the food landscapes of Lansing and East Lansing. Nevertheless, food insecurity has been an enduring concern for its residents. For instance, from 1945 onwards, auto workers at the Reo Motor Car Company included hunting in their workplace negotiations. Hunting was such an integral part of workers’ lives that the unions bargained with employers to grant time off to hunt; workers also lobbied their representatives to gain greater access to hunting grounds [58,59,121]. The workers’ campaigns suggest that access to food is more than access to neighborhood food retailers; access also involves recreational and subsistence activities, such as hunting and fishing.

Earlier food studies in East Lansing examined urban agriculture [122] and how seasonal variations affected walkability and food access [123]. Other researchers analyzed the availability of 94 food items in Lansing’s supermarkets, grocery stores, and convenience stores. The study focused on improving accuracy in mapping food outlets [124]. However, the study’s focus on a limited range of food outlets meant that it did not analyze most of the food venues in the local food environment.

We argue that researchers must examine a wide variety of food outlets to understand a locale’s food environment more fully. For instance, it behooves us to understand more about the distribution of fast-food outlets. This is crucial because, nationally, a third of adults eat fast food at least once per month [125]. Fast-food outlets are a critical segment of Michigan’s food landscape because 80% of the residents say they eat fast food at least once monthly, while 28% reported having it twice weekly. Sixty-four percent of the respondents said they consume fast food because it is quick and easily accessible [126].

Scholars also analyzed the relationship between eating at fast-food restaurants and neighborhood demographic characteristics. They found that 55% of the residents of Lansing’s poor neighborhoods, 26% of the area’s inner-suburban residents, and 13% of outer-suburban residents reported dining in fast-food restaurants monthly [127]. Researchers explain these patterns by arguing that fast-food restaurants are sometimes the only accessible food outlet for those without motorized transport. Such restaurants are often considered the convenient option when travel distance and foot accessibility [97,124,126,127], and cost [96,126] are essential to obtaining food.

Reed [128] studied Lansing’s food outlets and found that supermarkets were the most frequented; 63% of the respondents shopped in a supermarket at least once per week. Grocery stores and fast-food restaurants were also popular; 39% of the respondents patronized these venues weekly.

Other studies have explored physical access to food by examining the availability of transportation to help residents reach food venues [127]. Reed [127] reported that residents of metropolitan Lansing say that food retailers are not close to their primary residence. However, only seven percent of the respondents felt that lack of access was a serious problem. The researcher also found that 88% of the study participants used automobiles to get to the stores. However, 35% of the people shopping at liquor stores and 22% of those shopping at convenience stores were pedestrians. Reed also found that only 22% of the study respondents indicated that they did not do all or most of their food shopping near their primary residences. Reed’s study participants also reported visiting multiple locations when shopping for food, and 82% of them felt that food shopping was not a problem in their neighborhood. The usual travel time to get to various food outlet types ranged from 4.6 min to 13.5 min. Residents typically traveled about 10.3 min to get to a supermarket.

Seto & Ramankutty [129] analyze how transportation systems impact food access. Other researchers also examine how proximity to local stores and access to transportation influence residents’ ability to obtain desired food [96,97,130,131]. Studying only a locale’s supermarkets and large grocery stores can understate the contributions of ethnic restaurants and grocery stores to neighborhood food security. Local ethnic food stores provide foods that neighborhood residents want; hence, this is an important dimension of food access that must be considered [96,112]. Consequently, this study examines the prevalence of small and large food venues. Our approach builds on the work of researchers suggesting that small grocery stores can provide low-cost produce and food items some shoppers deem culturally desirable and necessary [112].

As mentioned above, East Lansing is the home to Michigan State University, a large public institution of higher education. Hence, a brief discussion of food insecurity amongst college students is appropriate here. Numerous college food insecurity studies have been conducted in the U.S. [132,133,134,135,136,137]. More specifically, several state and regional assessments have been conducted in Maryland and the Mid-Atlantic area [138,139] Illinois [140], Oregon [141], California and the western U.S. [142,143,144,145,146], and Georgia and the southeastern U.S. [147,148,149]. Research conducted at Michigan State University shows that in 2020, 39% of the students reported low or very low food security [46]. Student food insecurity is high at Michigan State, even though the first campus-based food bank in the U.S. was founded there in 1993 [150].

3. Methodology

3.1. Defining the Study Area

In keeping with the systems approach, we undertook a comprehensive study of food outlets in Lansing, East Lansing, and surrounding townships that sell, manufacture, grow, produce, distribute, process, aggregate, trade, or give away food. We extended the study area beyond the borders of the study cities to account for the edge effect. Researchers argue that urbanites do not limit themselves to their city boundaries when shopping for food; they commonly purchase food in nearby suburbs. Therefore, it is prudent to create a buffer around each city and collect and analyze data within it; this offers a more accurate assessment of food access than examining only the food outlets in a city [17,25,29,31,32,151].

Depending on the city’s size, the extent of urban sprawl, and customers’ shopping behavior, researchers use buffers as small as 4 kilometers or 2.5 miles [31] and as large as 16 km or about 10 miles [32] in food access studies. We used an eight-kilometer (roughly five-mile) buffer around the two cities to delimit our study area. Our analysis of food outlet locations showed that the number of food venues declined dramatically beyond this boundary. Using a buffer during a pandemic is salient, as many food retailers added home deliveries to their customer services [40]. Our buffer captured most food retailers in and around Lansing and East Lansing.

3.2. Data Collection and Sources

We collected data from September 2020 to June 2022. We developed a typology of 13 major categories and 57 sub-categories of food outlets [29,40] and studied Lansing, East Lansing, and surrounding townships to determine how many of each type were in the study area. We used the Food Marketing Index (FMI) to help us standardize definitions and classify grocery stores (see Appendix A). We adopt the definitions of full-service and fast-food restaurants used by Block et al. [152] and Taylor & Ard [17].

We used the index developed by Taylor et al. [40] as their typology includes many food outlet types often ignored in food access studies or not examined before (see Appendix A). We relied on Data Axle (formerly known as Reference USA), an information repository on U.S. and Canadian businesses, as our primary data source [153]. Rybarczyk et al. [31], Taylor & Ard [17], Liese et al. [115], Lisabeth et al. [154], and Raja et al. [155] also use Data Axle in their food studies.

We found food outlets by using the Standard Industrial Classification (SIC) division code. Other scholars analyzing food access utilize this technique [17,31,93,154,155]. We identified food venues from a list of 153 SIC codes. More in-depth data collection is necessary to find all the relevant food retailers since the SIC division codes identify only some of the pertinent food outlets. Consequently, we found additional food outlets using the major group, industry group, and industry codes as filters to help find other food establishments [29,40].

This approach helped us identify vendors (like mass merchandisers, supercenters, and variety stores) that sell food but are not classified as food stores in Data Axle. Finding these food outlets is essential, as many stores practice channel blurring and are significant food sources for some residents. Channel blurring is the process by which retailers expand their product lines to include the sale of food items. It occurs when retailers, such as pharmacies, dollar stores, or department stores, expand their product lines to include fresh produce, refrigerated and frozen items, and pantry staples [97]. Because food is one of many types of commodities sold, one will not find such food establishments by relying solely on the SIC division codes.

Downloaded data must be cleaned to remove duplicates and vendors that are not food related. Ergo, we ensured that each relevant vendor appeared only once in the database and deleted any retailer that was not a food purveyor from our downloads. Rybarczyk et al. [31] and Taylor & Ard [17] use these multi-pronged search and cleaning techniques to identify food outlets in Data Axle.

Information on food outlets was also collected from lists of local emergency food assistance programs; urban farms; community gardens; and stores that accept Women, Infants, and Children (WIC), and Electronic Benefits Transfer (EBT) cards. We also verified information using Google and Bing search engines and maps to identify addresses and latitude/longitude coordinates, business names, type of food outlet, cuisine served, phone number, year established, and the owner or manager’s demographic characteristics.

3.3. Spatial Mapping

ArcGIS Pro 10.8.1 was used to plot each food outlet’s latitude and longitude coordinates as a dot on the maps. We obtained the city boundaries from the U.S. Census Bureau’s [12,13,156] Open Data mapping tool, and then we extracted the shape files for Lansing, East Lansing, and the surrounding townships. The townships are – Lansing Charter Township, Dewitt Charter Township, Meridian Charter Township, Delhi Charter Township, Windsor Charter Township, Oneida Charter Township, Watertown Township, Bath Charter Township, Alaeidon Township, Delta Charter Township, Eagle Township, Dewitt Township, and Benton Township. Portions of the following townships were also included in the study area: Westphalia, Riley, Olive, Victor, Sciota, Woodhull, Williamstown, Wheatfield, Eaton, Eaton Rapids, Aurelius, Vervay, and Ingham.

We also merged neighborhood characteristics and 2020 census tract data with the spatial data. Others using this approach include Bell & Taylor [29], Andreyeva et al. [157], LeDoux & Vojnovic [21], Rose et al. [158], and Zenk et al. [25]. Food retailer information was added as a comma-separated value file. The NAD 1983 Michigan GeoRef was used to project the shapefiles onto the map [159].

We obtained the population size, racial characteristics, median household income, and percentage of residents with a high school education for each census tract from the 2020 census. We employed a technique used by Taylor et al. [40] that calculated the demographic data for census tracts that were only partially contained within the buffer. When census data are downloaded, there is an assumption that entire census tracts are fully contained within a city or county. However, this is not the case in our study area. Ergo, we clipped census tracts to correspond to the boundaries of our buffer. We then used proportional allocation to estimate the demographic factors we studied in the partial census tracts. Finally, we joined the census tract and food retailers’ shapefile to generate maps showing the location of food outlets in the census tracts. Rosencrants et al. [160] and Taylor et al. [40] take this approach also.

3.4. Statistical Analysis

We analyzed the demographic characteristics of the census tracts. First, we created a categorical variable to signify the racial composition of the tracts. The census tracts were divided into four groups: census tracts with 0–20% POC population were considered very low POC (VL-POC) tracts. The ones containing 20.01–40% POC residents were described as low POC (L-POC). The high POC (H-POC) tracts were ones with 40.01–60% Residents of Color. The last category, the very high POC (VH-POC) tracts, consists of census tracts where more than 60% of the residents are People of Color.

Given the non-normal distribution of the data, a Kruskal–Wallis H test supplemented by Dunn’s post hoc test for multiple comparisons was employed to determine whether there were statistically significant differences in the number and types of food outlets in the four census tract groups. Bonferroni correction for multiple comparisons was used (p-values reported in the results section take this correction into account).

Food access scholarship has employed several of the analyses used in our study. Diaz-Beltran et al. [161] used Kruskal–Wallis. The Kruskal–Wallis test is an omnibus procedure to identify the median difference. After the null hypothesis is rejected, multiple pairwise comparisons are conducted. To retain the rank sums from the Kruskal–Wallis test and to estimate a z-statistic similar to the exact rank sum statistic, Dunn’s test is considered a more appropriate procedure than the Mann–Whitney U test analysis followed by Dunn’s test with Bonferroni correction to analyze the effect of adults’ dietary autonomy when visiting fast-food restaurants [42]. Elbel et al. [162] also performed pairwise t-tests with the Bonferroni correction to compare the number of food outlet types between their created racial and income subgroups.

We conducted both Poisson and negative binomial regression analyses to assess the relationship between the number of food outlets in a census tract (dependent variable) and the independent variables—the racial composition of the census tract, population density (pop/km²), median household income, and percentage of the population with a high school education. The last three independent variables were continuous. The dispersion of the dependent variable determined whether we ran a Poisson or negative binomial regression. Poisson regression was used for food outlets that were not over-dispersed. Investigators such as Mundorf et al. [163], Rose et al. [158], Bodor et al. [104], and Moreland [94] utilize this approach. Negative binomial regression was used for outlets that exhibited overdispersion. Scholars such as Taylor et al. [40], Atkins & Gallop [164], Fine & Van Rooij [58], and Coxe et al. [165] apply this procedure.

Two models were analyzed. Model 1 (the crude model) examined only one independent variable—the racial characteristics of the census tracts. Model 2 was the adjusted model that included four independent variables—racial characteristics of the census tracts, population density (pop/km²), median household income, and educational attainment—in the analysis.

In the regression models, F is food outlet type; CTR is census tract race; I is the median household income in USD 1000; E is the percent of the population with a high school degree or equivalent; and PD is population density.

Model I:

log (F) = β₀ + β₁ × CTR

Model II:

log (F) = β₀ + β₁ × CTR + β₂ × I + β₃ × E + β₄ × PD

We checked for multicollinearity by assessing the variance inflation factor (VIF). The paper has a VIF cutoff of ≥2.5; Taylor et al. [40] and Johnston et al. [166] apply this threshold in their studies. We analyzed the sum of the food outlets and the categories of food sources in each census tract. However, we excluded some types of food outlets from the regression analysis if there were too few observations of particular food outlet types. We used IBM SPSS Statistics Version 28 to perform the analyses.

4. Results

4.1. The Lansing, East Lansing, and Suburban Food Landscape

We use the search techniques outlined above to identify the food outlets in the study area. Any closed food outlets were removed from the data set and tracked in a separate file; separate analyses will be conducted with these in the future. This article examines the 1647 open and operational food outlets identified in the study area.

Table 2. Food Outlets in Lansing, East Lansing, and Surrounding Townships, Michigan.

Food Outlet Type	Cities and Townships		Lansing		East Lansing		Surrounding Townships
	Frequency	Percent	Frequency	Percent	Frequency	Percent	Frequency	Percent
All food venues:	1647	100.0	579	100.0	220	100.0	848	100.0
Supermarkets and large grocery stores:	93	5.6	44	7.6	8	3.6	41	4.8
Traditional supermarkets and large groceries	36	2.2	19	3.3	3	1.4	14	1.7
Limited-assortment stores	26	1.6	19	3.3	1	0.5	6	0.7
Mass merchandisers	15	0.9	3	0.5	1	0.5	11	1.3
Supercenters	10	0.6	1	0.2	2	0.9	7	0.8
Fresh-format supermarkets	4	0.2	2	0.3	0	0.0	2	0.2
Wholesale clubs	2	0.1	0	0.0	1	0.5	1	0.1
Small groceries and convenience stores:	171	10.4	66	11.4	12	5.5	93	11.0
Gas stations with food	80	4.9	30	5.2	4	1.8	46	5.4
Liquor stores and party stores	43	2.6	19	3.3	2	0.9	22	2.6
Small groceries, convenience, and corner stores	48	2.9	17	2.9	6	2.7	25	2.9
Pharmacies, dollar, and variety stores:	119	7.2	45	7.8	11	5.0	63	7.4
Pharmacies or drug stores	73	4.4	22	3.8	11	5.0	40	4.7
Dollar stores and variety stores	46	2.8	23	4.0	0	0.0	23	2.7
Specialty food stores and vendors:	68	4.1	13	2.2	9	4.1	46	5.4
Bakeries	33	2.0	7	1.2	6	2.7	20	2.4
Ice cream parlors	21	1.3	4	0.7	3	1.4	14	1.7
Health food and nutrition supplements	6	0.4	1	0.2	0	0.0	5	0.6
Meat markets and delicatessens	6	0.4	1	0.2	0	0.0	5	0.6
Food cooperative	1	0.1	0	0.0	0	0.0	1	0.1
Gourmet shop	1	0.1	0	0.0	0	0.0	1	0.1
Restaurants and other food service:	772	46.9	224	38.7	107	48.6	441	52.0
Full-service restaurants	251	15.2	82	14.2	29	13.2	140	16.5
Fast-food restaurants	266	16.2	76	13.1	36	16.4	154	18.2
Takeout	74	4.5	26	4.5	6	2.7	42	5.0
Coffee, tea, and juice shops	73	4.4	14	2.4	20	9.1	39	4.6
Banquet halls and hotels	53	3.2	8	1.4	6	2.7	39	4.6
Bars and clubs	42	2.6	13	2.2	8	3.6	21	2.5
Caterers	12	0.7	5	0.9	2	0.9	5	0.6
Management	1	0.1	0	0.0	0	0.0	1	0.1
Farms, gardens, farmers’ markets, and produce vendors:	91	5.5	58	10.0	6	2.7	27	3.2
Urban farms and community gardens	65	3.9	50	8.6	5	2.3	10	1.2
Farmers’ markets and produce markets	24	1.5	7	1.2	1	0.5	16	1.9
Community-supported agriculture (CSA)	2	0.1	1	0.2	0	0.0	1	0.1
Food assistance:	56	3.4	37	6.4	4	1.8	15	1.8
Food pantries or soup kitchens	40	2.4	30	5.2	1	0.5	9	1.1
Food banks/distribution	15	0.9	7	1.2	2	0.9	6	0.7
Homeless shelter	1	0.1	0	0.0	1	0.5	0	0.0
Mobile food sources:	19	1.2	11	1.9	2	0.9	6	0.7
Mobile produce distributor	2	0.1	2	0.3	0	0.0	0	0.0
Food trucks	11	0.7	7	1.2	1	0.5	3	0.4
Mobile food distribution	6	0.4	2	0.3	1	0.5	3	0.4
Attractions and amusement parks:	9	0.5	1	0.2	1	0.5	7	0.8
Theater	2	0.1	1	0.2	1	0.5	0	0.0
Attractions and amusement parks	7	0.4	0	0.0	0	0.0	7	0.8
Social, religious, educational, and community services:	188	11.4	58	10.0	57	25.9	73	8.6
Associations	2	0.1	0	0.0	0	0.0	2	0.2
School cafeterias	48	2.9	25	4.3	13	5.9	10	1.2
Retirement communities and homes	28	1.7	5	0.9	3	1.4	20	2.4
Childcare	38	2.3	15	2.6	3	1.4	20	2.4
Religious institutions	10	0.6	4	0.7	1	0.5	5	0.6
Community centers	5	0.3	4	0.7	0	0.0	1	0.1
Service organizations	14	0.9	1	0.2	2	0.9	11	1.3
Youth organizations and centers	3	0.2	0	0.0	0	0.0	3	0.4
Social club	1	0.1	1	0.2	0	0.0	0	0.0
Group home	1	0.1	1	0.2	0	0.0	0	0.0
College and university food venues	38	2.3	2	0.3	35	15.9	1	0.1
University bakery	1	0.1	0	0.0	1	0.5	0	0.0
University cafés and coffee shops	4	0.2	0	0.0	4	1.8	0	0.0
University cafeterias or dining halls	13	0.8	1	0.2	12	5.5	0	0.0
University dairy products	2	0.1	0	0.0	2	0.9	0	0.0
University fast food restaurant	2	0.1	1	0.2	1	0.5	0	0.0
University food pantry or soup kitchen	1	0.1	0	0.0	1	0.5	0	0.0
University full service restaurant	1	0.1	0	0.0	1	0.5	0	0.0
University small groceries, corner, or convenience stores	14	0.9	0	0.0	13	5.9	1	0.1
Gyms and health centers:	22	1.3	3	0.5	2	0.9	17	2.0
Fitness centers, gyms, and health centers	18	1.1	0	0.0	2	0.9	16	1.9
Hospitals and medical centers	4	0.2	3	0.5	0	0.0	1	0.1
Supply chain:	38	2.3	19	3.3	1	0.5	18	2.1
Wholesalers	20	1.2	10	1.7	1	0.5	9	1.1
Manufacturers, processors	14	0.9	9	1.6	0	0.0	5	0.6
Distributors	4	0.2	0	0.0	0	0.0	4	0.5
Internet, online purchase, and delivery:	1	0.1	0	0.0	0	0.0	1	0.1
E-commerce, online	1	0.1	0	0.0	0	0.0	1	0.1

shows that the study area contained all 13 major categories of food outlets and 53 of the 57 sub-categories listed in Appendix A. The distribution of these outlets in and around the cities of Lansing and East Lansing is depicted in Figure 1. Almost half of the food outlets were in the two cities—579 (35.2%) were in Lansing, and 220 (13.4%) were in East Lansing. The remaining 848 (51.5%) outlets are in the surrounding townships.

Restaurants and other food services dominated the food environment. Social, religious, educational, and community service organizations were a distant second. The third-largest category was small grocery and convenience stores, while pharmacies, dollar, and variety stores were the fourth most frequently identified food venues. The categories containing supermarkets/large grocery stores and agricultural production (farms, community gardens, farmers’ markets, and produce vendors) round out the top six.

4.1.1. Supermarkets and Large Grocery Stores

The 93 supermarkets and large grocers in the study area represent only 5.6% of the food outlets studied. Lansing’s 44 supermarkets and large grocery stores comprise 7.6% of that city’s food environment. Only eight supermarkets and large grocery stores are in East Lansing; this accounts for 3.6% of the city’s food outlets. The 41 supermarkets and large grocery stores in the peri-urban area constituted 4.8% of the food landscape of the surrounding townships (Figure 2).

Lansing had most of the traditional supermarkets/large grocery stores and limited-assortment food stores, but the mass merchandisers and supercenters were in the surrounding townships. Despite being in peri-urban areas, the big-box stores were near each city.

4.1.2. Small Groceries and Convenience Stores

Small grocery and convenience stores were more prevalent than supermarkets and large grocery stores in the study area (Figure 3). A total of 171 small groceries and convenience stores were identified; these comprised 10.4% of the food environment. The small groceries and convenience stores made up roughly 11% of the food landscape of Lansing and the surrounding townships, but only 5.5% of East Lansing’s food outlets.

Twenty-five convenience stores, corner stores, and mini marts were found in the peri-urban area, and seventeen were in Lansing. Six more were found in East Lansing that were not university-based (the study found an additional fourteen campus-based mini marts and convenience stores—these are recorded under college and university food venues). Gas stations that sell food were the most common sub-category of the small grocery store sector; there were 80 such venues. Most gas stations (46) were in the suburbs, 30 were in Lansing, and only 4 were in East Lansing. There were 43 liquor and party stores in the study area. Twenty-two (51%) of them were in the suburbs. In Michigan, party stores are often liquor stores selling primarily alcohol. It should also be noted that alcohol can also be purchased in supermarkets, grocery stores, mass merchandisers, pharmacies, etc.

4.1.3. Pharmacies, Dollar, and Variety Stores

Food is regularly sold in Michigan’s pharmacies and drug stores. Our study identified 73 of these retailers that sold food; this amounts to 4.4% of the study area’s food outlets. There were 22 pharmacies and drugstores selling food in Lansing, 11 in East Lansing, and 40 in the surrounding townships.

Dollar stores and variety stores are also places where people frequently purchase food. We identified 46 dollar and variety stores in the study area. While East Lansing had none of these stores, half were found in Lansing, and the other half in the peri-urban area (Figure 4).

4.1.4. Specialty Food Stores and Vendors

Sixty-eight specialty food outlets were identified; most (forty-six) were in the surrounding townships. Thirteen were in Lansing, and nine were in East Lansing. Bakeries were the most common specialty food outlet, accounting for 48.5% of the category. Notably, the only specialty food stores in East Lansing were bakeries and ice cream parlors (Figure 5).

4.1.5. Restaurants and Other Food Services

Restaurants and other food services were the most abundant food retailers studied (Figure 6). Not all the restaurants and food services are visible in Figure 6 because of overlapping data points. In addition, one full-service restaurant and two fast-food restaurants not depicted here are shown on a later map containing university food outlets.

The 772 restaurants and food service entities constitute 46.9% of all food outlets in this study. Although restaurants comprised 38.7% of the total food outlets in Lansing and 48.6% in East Lansing, restaurants constituted 52% of the food outlets in the peri-urban area.

Most restaurants (441) were in the peri-urban area, while 224 were in Lansing, and 107 were in East Lansing. Fast-food (266) and full-service (251) establishments are the most common types of restaurants; they comprise two-thirds of the restaurants studied. We also identified takeout establishments and found 74 in the study area. Most takeouts (42) were in the peri-urban area, while 26 were in Lansing, and the remaining 6 were in East Lansing. The study area also had many coffee, tea, and juice shops. There were 73 in all. Lansing had the least (14), East Lansing had 20, and the surrounding townships had 39.

4.1.6. Urban Farms, Community Gardens, Farmers’ Markets, and Produce Vendors

The metropolitan area had a robust agricultural infrastructure. The percentage of agriculture-based food outlets is similar to that of supermarkets and large grocery stores. We identified 91 entities in this food category (Figure 7). Together, these comprised 5.5% of the total food outlets studied. There are 65 urban farms and community gardens in the study area; 50 are in Lansing, 10 are in the peri-urban area, and 5 are in East Lansing. Twenty-four farmers’ markets and produce vendors were identified; two-thirds were in the surrounding townships, while seven were in Lansing. Only one was found in East Lansing.

4.1.7. Emergency Food Assistance Organizations

Despite the robust food retail infrastructure in the study area, there is a strong demand for food that is given away at no cost. Hence, we found 56 emergency food assistance operations; this represents 3.4% of the food establishments examined (Figure 8). Food assistance organizations are mainly in Lansing; 37 are in this city. Four food assistance organizations are in East Lansing, and fifteen are in the peri-urban area. Food pantries and soup kitchens account for 71.4% of the entities in this category.

4.1.8. Mobile Food Sources

Mobile food services are increasingly popular, as they facilitate access to food (Figure 9). The location of each food truck is not visible on the map because several operate adjacent to each other. Consequently, some pinpoints overlap. The study identified 19 mobile food vendors, constituting 1.2% of the food outlets studied. Eleven were in Lansing, two were in East Lansing, and six were in surrounding townships. Food trucks were the most common type of mobile food vendor—57.9% (11) of the mobile types of food sources were food trucks.

4.1.9. Social, Religious, Educational, and Community Service Food Outlets

The study area contained 188 social, religious, educational, and community service organizations that sell, serve, or give away food (Figure 10). This category accounts for 11.4% of all food outlets, making it the second-largest category. School cafeterias (48), childcare facilities (38), and college and university food venues (38) were the most numerous subcategories. Lansing had 52% of the school cafeterias, but East Lansing had 92.1% of all the college and university food venues.

4.1.10. Supply Chain Vendors

The 38 supply chain food vendors identified represent 2.3% of the study area’s food outlets. Lansing has a robust supply chain sector; 19 are in that city. The supply chain establishments are arrayed along Business 96, a major north–south thoroughfare that dissects the city. Only 1 supply chain establishment (a wholesaler) was found in East Lansing, while 18 are in the peri-urban area. Wholesalers were the most common of the supply chain vendors studied. They accounted for 52.6% of this category of vendors (Figure 11).

4.1.11. Miscellaneous Food Sources

We also identified 32 miscellaneous types of institutions that sold food: attractions (such as amusement parks, fairs, or circuses), fitness and health centers, hospitals and medical centers, and e-commerce or online vendors. Only e-commerce or online vendors based in the study area were included in this study. All locations are not visible for fitness centers and health centers because four outlets have the same address (Figure 12). The study area had 9 attractions; 22 fitness centers, health centers, and hospitals; and 1 e-commerce vendor.

4.2. Race, Population Size, Socio-Economic Indicators, and Food Landscape

4.2.1. Types of Census Tracts Studied

This paper seeks to understand the connection between census tract characteristics and the distribution of food outlets. The study area contained 115 census tracts. The one census tract with no residents is in Lansing and is depicted in light yellow. It is excluded from this portion of the analysis [6]. See Figure 13.

The proportional allocation function of ArcGIS Pro 10.8.1 was used to calculate the population totals. The population totals are derived from the American Community Survey’s 5-year estimates for 2020. Analyzing the census tracts in the study area is complicated because, as Figure 13 shows, several census tracts straddle the boundaries of each city and the boundaries between the cities and the surrounding townships. In all, 27 census tracts span city and township boundaries. Despite the overlaps, all the census tracts were fully contained within the four counties (Clinton, Eaton, Ingham, and Shiawassee) the study area encompassed.

The 114 tracts studied contained 358,185 residents. We used a categorical variable to describe the racial composition of the census tracts. Fifty-two tracts containing 156,696 residents were classified as VL-POC. Overwhelmingly, the VL-POC census tracts were located outside of the city boundaries. However, 11 of Lansing’s and 6 of East Lansing’s census tracts were classified as VL-POC tracts (

Table 3. Demographic Characteristics of Census Tracts.

Census Tract Racial Characteristics	Number of Census Tracts	Demographic Characteristics
		Population Size	Number of People of Color in Tract a	Population Density, km2	Household Income	Percentage Completed High School	Percentage Living in Poverty
Very Low People of Color, VL-POC (0–20%)	52	156,696	17,697
Minimum		170	5	23.03	$20,373.00	85.08	1.90
Maximum		7024	1163	6216.30	$111,285.00	99.77	67.30
Median		3009	313	298.42	$67,763.00	95.75	7.00
Standard Deviation		1752	261	985.24	$20,190.84	2.74	11.98
Low People of Color, L-POC (20.01–40%)	38	123,396	36,717
Minimum		807	216	39.99	$18,897.00	57.14	2.20
Maximum		6401	1795	22040.22	$130,598.00	100.00	67.80
Median		3248	924	1312.49	$56,801.00	96.36	13.35
Standard Deviation		1156	364	4757.93	$25,825.83	7.79	14.10
High People of Color, H-POC (40.01–60%)	19	67,240	31,604
Minimum		2018	836	91.68	$12,125.00	77.48	2.00
Maximum		5654	2655	3698.28	$80,850.00	99.19	81.20
Median		3568	1588	1236.14	$41,555.00	90.46	20.90
Standard Deviation		1049	538	891.95	$13,313.98	6.32	17.89
Very High People of Color, VH-POC (Over 60%)	5	10,852	7174
Minimum		52	47	10.51	$27,264.00	82.94	0.00
Maximum		3992	2567	2346.92	$48,795.00	100.00	45.20
Median		2187	1357	2165.53	$31,513.00	89.35	22.10
Standard Deviation		1716	1105	980.48	$10,536.07	7.60	16.06

Notes: ^a The size of the People of Color population was averaged across all tracts in a category. Statistical tests were not performed because the differences were by design. That is, a census tract was designated a very low People of Color tract if 0–20% of the residents were People of Color. A tract was designated a low People of Color tract if 20.1–40% of the inhabitants were People of Color. Census tracts wherein 40.01–60% of the residents are People of Color were categorized as high People of Color tracts. If more than 60% of the inhabitants of a tract were People of Color, those tracts were described as very high People of Color tracts. Source: [6]

).

The 38 L-POC tracts had 123,396 residents; these tracts were more evenly distributed between the cities and the suburbs. The 19 H-POC tracts had 67,240 residents; these tracts were most often found in Lansing. Lastly, the five VH-POC tracts contained 10,852 people; four were in Lansing, while one was in East Lansing. Although the VL-POC census tracts have substantially more residents than the other types of tracts, the VL-POC tracts have much lower population densities than the three remaining categories of tracts. Table 3 shows that the VL-POC tracts have a much lower median population than the remaining three categories of census tracts. The VL-POC tracts are more sparsely populated because they are dispersed over semi-rural townships.

There are substantial income differences among the census tracts, and income is associated with the racial characteristics of the tracts. A weighted median per census tract racial category was calculated for the median household income reported in Table 3. Within each census tract racial category, there are varying numbers of census tracts, each with its own median household income. Weighted medians help to account for variations in population sizes among different categories or groups within the table. This method ensures that our analysis accurately reflects the central tendency of the data and provides a more representative measure of demographic characteristics. This is essential when the population size of the groups varies significantly.

The data shows that the median household income decreases as the percentage of POC in the census tracts increases. While the median household income is USD 67,763 in VL-POC tracts, it is USD 31,513 in VH-POC tracts. The reverse is true for poverty. The median poverty rate rises as the percentage of POC in the tract increases. There is no strong relationship between educational attainment and the census tracts people live in. However, the percentage of residents completing high school is slightly lower in census tracts where more than 40% of the residents are POC than in tracts where 40% or less of the inhabitants are POC.

4.2.2. Comparing the Distribution of Food Outlets in Census Tracts

There are noticeable differences between the four census tract groupings (see

Table 4. Number of Residents, Food Outlet Types, and Racial Composition of Census Tracts in Lansing, East Lansing, and Surrounding Townships.

Major Food Outlet Category	Subcategory of Food Outlets b	Census Tract Racial Characteristics
		Total Population a					0–20.00% People of Color (VL-POC)					20.01–40.00% People of Color (L-POC)					40.01–60.00% People of Color (H-POC)					60.01% or More People of Color (VH-POC)
		Combined Number of Residents in Census Tracts with Outlets	Number of Each Type of Food Outlet c	Number of Census Tracts Food Outlets are In	Number of Food Outlets Per 1000 Persons	Percent of Food Outlet Contained in Census Tracts	Combined Number of Residents in Census Tracts with Outlets	Number of Each Type of Food Outlet	Number of Census Tracts Food Outlets are In	Number of Food Outlets Per 1000 Persons	Percent of Food Outlet Contained in Census Tracts	Combined Number of Residents in Census Tracts with Outlets	Number of Each Type of Food Outlet	Number of Census Tracts Food Outlets are In	Number of Food Outlets Per 1000 Persons	Percent of Food Outlet Contained in Census Tracts	Combined Number of Residents in Census Tracts with Outlets	Number of Each Type of Food Outlet	Number of Census Tracts Food Outlets are In	Number of Food Outlets Per 1000 Persons	Percent of Food Outlet Contained in Census Tracts	Combined Number of Residents in Census Tracts with Outlets	Number of Each Type of Food Outlet	Number of Census Tracts Food Outlets are In	Number of Food Outlets Per 1000 Persons	Percent of Food Outlet Contained in Census Tracts
Total		358,185	1,499	114	4.18	100.00	156,697	578	43	3.69	38.56	123,396	454	32	3.68	30.29	67,240	450	19	6.69	30.02	10,852	17	5	1.57	1.13
Supermarkets and large grocery stores	Traditional supermarkets	111,606	36	31	0.32	100.00	23,277	7	7	0.30	19.44	52,542	17	14	0.32	47.22	35,787	12	10	0.34	33.33	0	0	0	0.00	0.00
	Limited-assortment stores	68,122	26	19	0.38	100.00	3545	2	1	0.56	7.69	23,812	6	6	0.25	23.08	38,578	17	11	0.44	65.38	2187	1	1	0.46	3.85
	Mass merchandisers and supercenters	63,807	25	16	0.39	100.00	25,859	8	6	0.31	32.00	17,604	4	4	0.23	16.00	20,344	13	6	0.64	52.00	0	0	0	0.00	0.00
Small groceries and convenience stores	Gas stations with food	182,439	79	50	0.43	100.00	76,019	32	22	0.42	40.51	51,751	19	13	0.37	24.05	50,956	27	14	0.53	34.18	3713	1	1	0.27	1.27
	Liquor stores and party stores	139,015	43	36	0.31	100.00	62,444	18	16	0.29	41.86	36,284	11	10	0.30	25.58	32,582	12	8	0.37	27.91	7705	2	2	0.26	4.65
	Small groceries, convenience, and corner stores	131,306	47	36	0.36	100.00	46,956	15	13	0.32	31.91	47,673	16	13	0.34	34.04	36,677	16	10	0.44	34.04	0	0	0	0.00	0.00
Pharmacies, dollar, and variety stores	Pharmacies or drug stores	154,917	73	41	0.47	100.00	79,309	31	20	0.39	42.47	46,648	21	13	0.45	28.77	28,960	21	8	0.73	28.77	0	0	0	0.00	0.00
	Dollar stores and variety stores	136,918	46	37	0.34	100.00	45,017	13	12	0.29	28.26	42,168	14	12	0.33	30.43	47,546	18	12	0.38	39.13	2187	1	1	0.46	2.17
Specialty food stores and vendors	Bakeries	99,385	33	26	0.33	100.00	61,497	19	15	0.31	57.58	26,059	7	7	0.27	21.21	11,829	7	4	0.59	21.21	0	0	0	0.00	0.00
	Ice cream parlors	71,481	21	18	0.29	100.00	33,776	11	8	0.33	52.38	21,638	6	6	0.28	28.57	16,067	4	4	0.25	19.05	0	0	0	0.00	0.00
Restaurants and other food service	Full-service restaurants	264,622	250	75	0.94	100.00	121,187	91	34	0.75	36.40	82,460	91	23	1.10	36.40	57,210	66	16	1.15	26.40	3765	2	2	0.53	0.80
	Fast-food restaurants	235,271	266	64	1.13	100.00	98,564	106	25	1.08	39.85	76,169	71	22	0.93	26.69	60,538	89	17	1.47	33.46	0	0	0	0.00	0.00
	Takeout	156,973	74	42	0.47	100.00	76,480	35	20	0.46	47.30	44,345	20	12	0.45	27.03	29,970	17	8	0.57	22.97	6179	2	2	0.32	2.70
	Coffee, tea, and juice shops	158,281	73	42	0.46	100.00	80,237	33	20	0.41	45.21	50,332	24	14	0.48	32.88	27,712	16	8	0.58	21.92	0	0	0	0.00	0.00
	Banquet halls and hotels	117,126	53	31	0.45	100.00	59,637	24	15	0.40	45.28	36,042	22	10	0.61	41.51	21,447	7	6	0.33	13.21	0	0	0	0.00	0.00
	Bars and clubs	97,536	42	27	0.43	100.00	58,310	24	15	0.41	57.14	28,106	12	8	0.43	28.57	11,068	5	3	0.45	11.90	52	1	1	19.23	2.38
Farms, gardens, farmers’ markets, and produce vendors	Urban farms and community gardens	118,412	64	35	0.54	100.00	26,219	10	8	0.38	15.63	40,089	19	12	0.47	29.69	42,212	32	12	0.76	50.00	9892	3	3	0.30	4.69
	Farmers’ markets and produce markets	79,396	24	21	0.30	100.00	49,369	16	13	0.32	66.67	16,364	4	4	0.24	16.67	13,663	4	4	0.29	16.67	0	0	0	0.00	0.00
Food assistance	Food pantries or soup kitchens	83,649	40	24	0.48	100.00	12,416	5	3	0.40	12.50	35,457	17	11	0.48	42.50	33,589	17	17	0.51	42.50	2187	1	1	0.46	2.50
	Food banks/distribution	51,981	15	14	0.29	100.00	25,907	7	6	0.27	46.67	2467	1	1	0.41	6.67	19,615	6	6	0.31	40.00	3992	1	1	0.25	6.67
Social, religious, educational, and community services	School cafeterias	124,190	47	32	0.38	100.00	36,022	16	10	0.44	34.04	37,175	13	9	0.35	27.66	50,993	18	13	0.35	38.30	0	0	0	0.00	0.00
	Retirement communities and homes	85,914	28	24	0.33	100.00	68,219	19	16	0.28	67.86	11,744	4	4	0.34	14.29	15,951	5	4	0.31	17.86	0	0	0	0.00	0.00
	Childcare	104,680	38	29	0.36	100.00	46,531	15	11	0.32	39.47	32,523	15	10	0.46	39.47	20,726	6	6	0.29	15.79	4900	2	2	0.41	5.26
Gyms and health centers	Fitness centers, gyms, and health centers	47,646	18	13	0.38	100.00	37,777	11	10	0.29	61.11	6912	6	2	0.87	33.33	2957	1	1	0.34	5.56	0	0	0	0.00	0.00
Supply chain	Wholesalers	60,376	20	16	0.33	100.00	22,064	5	5	0.23	25.00	17,292	5	5	0.29	25.00	21,020	10	6	0.48	50.00	0	0	0	0.00	0.00
	Manufacturers, processors, and distributors	49,879	18	14	0.36	100.00	19,123	5	5	0.26	27.78	19,753	9	6	0.46	50.00	11,003	4	3	0.36	22.22	0	0	0	0.00	0.00
Summary for Census Tract Groupings
Percent of the population		100.00					43.74					34.45					18.77					3.03
Percent of the census tracts				100.00					45.61					33.33					16.67					4.39

Notes: ^a The columns with number of residents will sum to more than population in the totals row because several types of food outlets are located in a given census tract. ^b The columns with number of census tracts food outlets are in will exceed the number of census tracts displayed in the totals row because some census tracts contain more than one food outlet of a particular type (for instance, some census tracts contain more than one supermarket). ^c The count of the food outlets is lower than that reported in Table 2 because one census tract is not analyzed in this table. The excluded census tract has no residents, however, it contained food outlets. Those food outlets in that tract are excluded from this analysis.

). A summary of the grouped characteristics shows that the VL-POC census tracts contain the most significant number of food outlets (578 in all). These tracts contained 43.74% of the study area’s population, almost 46% of the census tracts, and 38.6% of the food outlets. The VL-POC tracts contained between 40% and 49% of seven types of food outlets. These are takeouts—47.3%; food banks/food distribution—46.7%; banquet halls and hotels—45.3%; coffee, tea, and juice shops—45.2%; pharmacies and drug stores—42.5%; liquor stores and party stores—41.9%; and gas stations with food—40.5%.

The VL-POC tracts also had 67.9% of all the retirement communities and homes; two-thirds of the farmers’ markets and produce markets; 61.1% of the fitness centers, gyms, and health centers; 57.6% of the bakeries; 57.1% of the bars and clubs; and 52.4% of the ice cream parlors.

The L-POC tracts contain 34.5% of the study area’s residents, a third of the census tracts, and 30.3% of the food outlets. The tracts contain 454 food venues—the second-largest number of outlets studied. Half of the wholesalers, processors, and distributors are in these tracts. In addition, the L-POC tracts contain the largest number of traditional supermarkets and large grocery stores (17). In other words, the tracts contain 47.2% of traditional supermarkets/large grocery stores. They also have 42.5% of the food pantries or soup kitchens and 41.5% of the banquet halls and hotels.

The H-POC census tracts have 18.8% of the population in the study area. They have one-sixth of the census tracts and 30% of the food outlets. There are 450 food outlets in the H-POC tracts. These tracts are well endowed with traditional supermarkets (12 or 33.3%), limited-assortment stores (17 or 65.4%), and mass merchandisers and supercenters (13 or 52%). The H-POC census tracts also have half of the urban farms and community gardens and half of the wholesalers. The tracts also contain 42.5% of the food pantries or soup kitchens and 40% of the food banks/distribution centers.

The analysis of the food outlet distribution in the VH-POC census tracts reveals that these tracts are much more under-resourced than other categories of tracts. The VH-POC tracts host only 11 of the 26 food outlet types analyzed in this section of the paper. For instance, there are no traditional supermarkets or large grocery stores; mass merchandisers or supercenters; or small groceries, convenience, and corner stores in the VH-POC tracts, despite having 10,852 residents. Hence, the VH-POC tracts have 3% of the residents in the study area, 4.4% of the census tracts, but only 1.1% of the food outlets. Seventeen of the food outlets in the VH-POC tracts are included in this portion of the analysis.

The VH-POC tracts have a grand total of 21 food outlets. The four not included in this portion of the analysis are one religious institution, one food truck, one mobile food distribution entity, and one meat/delicatessen venue. All of these were in food categories with cell counts that did not meet our cut-off threshold for multivariate analyses.

4.2.3. Racial Characteristics of Census Tracts and Food Outlets Per Thousand Persons

We analyzed food categories with cell counts of 15 or more—a count high enough to withstand multivariate analyses. Hence, 26 subcategories were analyzed. Eight food outlets were located in the census tract with no residents and are not included in the analysis. The total number of outlets analyzed in this portion of the paper is 1499.

There are 4.18 food outlets per 1000 people across the study area. VL-POC and L-POC census tracts have similar numbers of food outlets per 1000 residents. There are 3.69 food outlets per 1000 in the VL-POC tracts, while it is 3.68 per thousand in the L-POC tracts. The count jumps significantly in H-POC tracts to 6.69 food outlets per 1000. However, the number of food outlets per 1000 persons drops precipitously in the VH-POC census tracts; it is 1.57 food outlets per 1000 persons (see Table 4).

An assessment of supermarkets and large grocery stores shows that there are 0.32 per 1000 inhabitants of these food outlets in the study area. The rate of occurrence of supermarkets and large grocery stores is similar for three of the racial categories. It is 0.30 for VL-POC census tracts, 0.32 for L-POC tracts, and 0.34 for H-POC tracts. However, the pattern changes for VH-POC tracts. In these census tracts, where more than 60 percent of the residents are People of Color, the rate of occurrence is zero. Unlike the three other tracts, the VH-POC tracts also had a zero rate of occurrence for mass merchandisers and supercenters.

There was not a discernible pattern in the distribution of limited-assortment stores. The VL-POC had the highest per-thousand rate at 0.56 per thousand people. The VH-POC and H-POC tracts were in the middle, with rates of 0.46 and 0.44 per 1000 persons, respectively. The L-POC tracts had the lowest rate of 0.25 per 1000 persons.

How do the different groups of census tracts fare when small groceries and convenience stores are assessed? Contrary to expectations, the VH-POC census tracts had a zero rate of occurrence for small grocery stores, corner stores, and mini marts. The three remaining groups of tracts had occurrence rates of 0.32 to 0.44 per thousand residents. The VH-POC tracts continue to defy the findings and predictions of earlier food studies by recording the lowest rate of occurrence of liquor and party stores, as well as gas stations selling food, in the four types of census tract groups examined. So, liquor and party stores occurred at the rate of 0.26 per thousand in VH-POC tracts, while gas stations selling food occurred at the rate of 0.27 per thousand. In comparison, these two categories of food outlets occurred at rates of 0.29 or higher in all the remaining groups of census tracts.

While supermarkets might shun VH-POC census tracts, dollar stores and variety stores do not. Dollar stores and variety stores typically carry a limited line of processed, canned, bottled, bagged, and boxed groceries. Although the VH-POC tract has no pharmacies or drug stores that sell food, the per-1000-person rate for the dollar and variety store category is 0.46. The rate for dollar and variety stores in the other categories of census tracts ranges from 0.29 to 0.38 per 1000.

The VH-POC census tracts have a much lower rate of full-service restaurants per 1000 persons than other tracts. While other tracts have between 0.75 and 1.15 full-service restaurants per 1000 persons, the VH-POC tracts have only 0.53 restaurants per 1000 residents. Once again, the VH-POC tracts defy predictions; they have a zero count for fast-food restaurants. The remaining three types of census tracts have between 0.93 and 1.47 fast-food restaurants per 1000 residents. The VH-POC tracts also have a much lower rate of takeout restaurants than the other three categories of census tracts.

The H-POC census tracts have a significantly higher occurrence of urban farms and community gardens than the remaining groups of tracts. Urban farms and community gardens occurred at the rate of 0.76 per 1000 residents in H-POC tracts. In contrast, the rate of occurrence was 0.47 in L-POC tracts, 0.38 in VL-POC tracts, and 0.30 in VH-POC tracts. While there were zero occurrences of farmers’ markets and produce vendors in VH-POC tracts, the range in other tracts was 0.24 to 0.32 per 1000 persons.

Emergency food assistance establishments are crucial food sources in many urban areas. Food pantries and soup kitchens occur at similar rates throughout the study area (0.48 per 1000 persons). They range from a low of 0.40 per 1000 in VL-POC tracts to a high of 0.51 per 1000 in H-POC tracts. On the flip side, the VH-POC and VL-POC tracts had a much lower occurrence of food banks/distribution than the two remaining tracts. Hence, food banks/distribution occurred at the rate of 0.25 per 1000 residents in VH-POC tracts and 0.27 per 1000 in VL-POC tracts. It was 0.31 per 1000 in H-POC tracts and 0.41 per 1000 residents in L-POC census tracts.

4.2.4. Extremely Low-Access Census Tracts

The study identified a phenomenon not reported in earlier studies—ELA census tracts (see Figure 14). We identified 11 census tracts that had no food outlets in any of them. Of these ELAs, nine were VL-POC tracts (with 0–20% People of Color), and two were L-POC tracts (with 20.01–40% People of Color inhabitants). Ten of the ELA tracts were in the semi-rural periphery of the study area, and one was in Lansing.

4.2.5. Regression Analyses and Tests for Significance

Kruskal–Wallis, Dunn’s, and Bonferroni Correction

We performed a Kruskal–Wallis test to examine whether differences in the number of food outlet types in the four categories of census tracts were significant (

Table 5. Comparison of Food Outlets by Census Tract Racial Characteristics.

Food Outlet Type	Kruskal-Wallis a		Dunn’s Multiple Comparison Test b,d
			VL-POC & L-POC	VL-POC & H-POC	VL-POC & VH-POC	L-POC & H-POC	L-POC & VH-POC	H-POC & VH-POC
	H	p-Value c	p-Value	p-Value	p-Value	p-Value	p-Value	p-Value
Total	16.28	0.001 **	1.000	0.003 *	0.773	0.035 *	0.418	0.007 **
Traditional supermarkets and large grocery stores	15.378	0.002 **	0.073	0.005 **	1.000	1.000	0.500	0.107
Limited-assortment stores	31.906	0.000*	0.636	0.000 **	1.000	0.000 **	1.000	0.210
Mass merchandisers and supercenters	6.614	0.085
Gas stations with food	12.523	0.006 **	1.000	0.029 *	1.000	0.007 **	1.000	0.089
Liquor stores and party stores	2.251	0.522
Small groceries, convenience, and corner stores	8.476	0.037 *	1.000	0.080	1.000	0.614	0.796	0.117
Pharmacies or drug stores	3.878	0.275
Dollar stores and variety stores	13.152	0.004 **	1.000	0.002 *	1.000	0.039 *	1.000	0.249
Bakeries	3.089	0.378
Ice cream parlors	1.244	0.743
Full-service restaurants	7.984	0.046 *	1.000	0.145	1.000	0.506	0.701	0.086
Fast-food restaurants	17.995	0.000 **	1.000	0.003 **	0.396	0.020 *	0.250	0.002 **
Takeout	1.295	0.730
Coffee, tea, and juice shops	3.467	0.325
Banquet halls and hotels	1.928	0.587
Bars and clubs	1.58	0.664
Urban farms and community gardens	18.984	0.000 **	0.542	0.000 **	0.503	0.037 *	1.000	1.000
Farmers’ markets and produce markets	4.497	0.213
Food pantries or soup kitchens	16.847	0.001 **	0.057	0.001 **	1.000	0.514	1.000	0.895
Food banks/distribution	9.865	0.020 *	1.000	0.159	1.000	0.012 **	1.000	1.000
School cafeterias	18.772	0.000 **	1.000	0.000 **	1.000	0.004 **	1.000	0.018 **
Retirement communities and homes	6.865	0.076
Childcare	1.07	0.784
Fitness centers, gyms, and health centers	5.439	0.142
Wholesalers	7.053	0.070
Manufacturers, processors, and distributors	1.779	0.620

Notes: ^a Pairwise comparisons were computed for statistically significant Kruskal-Wallis results. ^b Adjustment for multiple comparisons: Bonferroni ^c Significance levels: * p-values <α = 0.05, ** p-values <α = 0.01 ^d No value indicates no pairwise comparison was conducted.

). The H values were statistically significant in the 12 food outlet types tested. The overall number of food outlets (H = 16.28, p = 0.001) was significant. The most significant value in the large grocery store sector was the limited-assortment stores (H = 31.91, p = 000). The traditional supermarkets/large grocery stores were also significant (H = 15.38, p = 0.002). The dollar and variety stores (H = 13.15, p = 0.004) had the highest significance among small food stores. This was followed by gas stations that sell food (H = 12.52, p = 0.006) and small groceries, convenience, and corner stores (H = 8.45, p = 0.037).

In the restaurant sector, fast-food restaurants (H = 17.99, p = 0.000) were more significant than full-service restaurants (H = 7.98, p = 0.046). There were significant H scores for urban farms/community gardens, food pantries/soup kitchens, and food banks. The urban farms and community gardens had an H score of 18.98 (p = 0.000). The food pantries and soup kitchens (H = 16.85, p = 0.001) and food banks/distribution (H = 9.87, p = 0.020) had high scores, too. School cafeterias (H = 18.77, p = 0.000) were also significant.

Further tests were performed to determine which categories of census tracts differed from each other. This was achieved by coupling each significant Kruskal–Wallis test with Dunn’s test with a Bonferroni correction. No significant differences were evident when comparing the VL-POC and L-POC census tracts. The same was true for comparing the VL-POC and the VH-POC—no significant results were obtained. One other comparison produced no significant results—L-POC and VH-POC census tracts.

Significant results were obtained for some food outlet types in comparisons between VL-POC and H-POC; nine were significant. They are limited-assortment stores (p = 0.000) and traditional supermarkets/large grocery stores (p = 0.005). The dollar and variety stores had a p-value of 0.002, and the gas stations selling food had a p-value of 0.029. The only significant restaurant type was fast-food restaurants (p = 0.003). The comparison of these two types of census tracts also revealed significant differences in urban farms/community gardens (p = 0.000), food pantries/soup kitchens (p = 0.001), and school cafeterias (p = 0.000).

The analysis also uncovered eight significant differences between the L-POC and H-POC tracts. Significant differences were observed in the overall number of food outlets (p = 0.035). There were also significant differences in the limited-assortment stores (p = 0.000), gas stations that sell food (p = 0.007), and dollar and variety stores (p = 0.039). The difference in the distribution of fast-food restaurants (p = 0.020) and urban farms/community gardens (p = 0.037) was also significant. There were also significant differences in the food banks/distribution (p = 0.012) and school cafeterias (p = 0.004).

Three significant differences were observed in the comparisons between the H-POC and VH-POC tracts. These are the overall food outlet count (p = 0.007), the number of fast-food restaurants (p = 0.002), and the school cafeterias (p = 0.018).

Poisson and Negative Binomial Analyses

We conducted Poisson and negative binomial regressions to determine the likelihood of observing an additional food outlet in census tracts with different racial characteristics. Each model controlled for population density, the median income, and the percentage of residents over 25 who have obtained at least a high school education. The reference group was VL-POC census tracts (

Table 6. Incidence Rate Ratios for Full Model Showing Census Tract Racial Characteristics.

Food Outlet Type	Census Tract Racial Characteristics
	20.01–40.00% People of Color (L-POC)				40.01–60.00% People of Color (H-POC)				60.01% or More People of Color (VH-POC) f
	Incidence Rate Ratio (IRR) a,b	95% Confidence Interval (CI)		p-Value g	Incidence Rate Ratio (IRR)	95% Confidence Interval (CI)		p-Value	Incidence Rate Ratio (IRR)	95% Confidence Interval (CI)		p-Value
		Lower	Upper			Lower	Upper			Lower	Upper
e Total number of food outlets	1.105	0.705	1.732	0.663	1.870	1.044	3.348	0.035 *	0.284	0.096	0.838	0.023 *
d Traditional supermarkets	3.655	1.452	9.203	0.006**	4.830	1.726	13.515	0.003 **
d Limited-assortment stores	4.025	0.743	21.796	0.106	17.320	3.525	85.100	0.000 **	2.805	0.181	43.499	0.461
e Mass merchandisers and supercenters	0.916	0.230	3.654	0.901	7.787	2.219	27.321	0.001 **
d Gas stations with food	0.941	0.506	1.749	0.848	2.164	1.177	3.979	0.013 **	0.238	0.029	1.942	0.180
d Liquor stores and party stores	0.616	0.263	1.442	0.264	1.071	0.468	2.450	0.871	0.491	0.096	2.521	0.394
d Small groceries, convenience, and corner stores	1.392	0.639	3.031	0.404	2.266	1.012	5.074	0.047 *
e Pharmacies or drug stores	0.882	0.417	1.868	0.744	1.485	0.648	3.403	0.350
d Dollar stores and variety stores	1.905	0.834	4.354	0.126	4.015	1.734	9.299	0.001 **	0.731	0.080	6.657	0.781
d Bakeries	0.467	0.180	1.212	0.118	1.249	0.464	3.362	0.660
d Ice cream parlors	0.942	0.321	2.764	0.914	1.168	0.319	4.268	0.815
e Full-service restaurants	1.455	0.859	2.464	0.163	1.934	0.978	3.826	0.058 *	0.148	0.023	0.931	0.042 *
e Fast-food restaurants	0.989	0.524	1.865	0.972	2.249	1.017	4.975	0.045 *
e Takeout	0.703	0.331	1.493	0.359	0.905	0.371	2.212	0.828	0.331	0.051	2.150	0.247
e Coffee, tea, and juice shops	0.960	0.476	1.936	0.908	1.168	0.500	2.728	0.720
e Banquet halls and hotels	1.475	0.682	3.192	0.324	0.794	0.267	2.355	0.677
e Bars and clubs	0.582	0.242	1.403	0.228	0.403	0.119	1.363	0.144	0.133	0.010	1.796	0.129
e Urban farms and community gardens	2.084	0.782	5.555	0.142	4.810	1.733	13.351	0.003 **	1.729	0.308	9.698	0.534
d Farmers’ markets and produce markets	0.348	0.106	1.143	0.082	0.618	0.178	2.143	0.448
d Food pantries or soup kitchens	4.854	1.708	13.799	0.003**	8.177	2.770	24.136	0.000 **	1.733	0.183	16.447	0.632
d Food banks/distribution	0.144	0.014	1.487	0.104	1.709	0.435	6.712	0.442	0.920	0.082	10.348	0.946
d School cafeterias	1.131	0.515	2.486	0.759	2.912	1.322	6.415	0.008 **
d Retirement communities and homes	0.361	0.114	1.144	0.083	0.857	0.279	2.628	0.787
e Childcare	1.474	0.604	3.596	0.394	1.002	0.305	3.287	0.998	1.266	0.197	8.160	0.804
e Fitness centers, gyms, and health centers	0.584	0.144	2.361	0.450	0.235	0.024	2.289	0.212
d Wholesalers	1.710	0.419	6.988	0.455	4.860	1.375	17.177	0.014 **
e Manufacturers, processors, and distributors	2.586	0.681	9.816	0.163	1.710	0.339	8.638	0.516

Notes: ^a Incident Rate Ratio values of less than 1 indicate that there are fewer stores than in the reference group. The reference group is the very low People of Color census tracts (0–20% People of Color). ^b Each row represents a separate model that adjusted for population density, median household income, and the percent of the population over 25 that has at least a high school education. ^c Model for the following food outlets: bakeries; bars and clubs; childcare; coffee, tea, and juice shops; manufacturers, processors, and distributors; farmers’ markets and produce markets; food bank/distribution; ice cream parlors; liquor stores and party stores; pharmacies and drug stores; and takeout was not significant, p-values reported as needed. ^d Indicates models that followed a poisson regression. ^e Indicates models that followed a non-binomial regression. ^f No value indicates that outlet type was not located in census tract racial category. ^g Significance levels: * p-values < than α= 0.05, ** p-values < than α= 0.01.

).

Several relationships were statistically significant. Only two of the comparisons between VL-POC tracts and L-POC tracts were significant. These were the differences between traditional supermarkets/large grocery stores and the distribution of food pantries or soup kitchens. The findings show that L-POC tracts were 3.66 times more likely (IRR 3.655; 95% CI = 1.452, 9.203) to have an additional traditional supermarket or large grocery store than VL-POC tracts. The comparison of these two types of tracts was even more significant, as L-POC tracts were 4.85 times more likely to have an additional food pantry or soup kitchen than VL-POC tracts (IRR 4.854; 95% CI = 1.708, 13.799).

The comparison of the VL-POC tracts and the H-POC tracts yielded the most numerous significant results; 13 of the 27 comparisons of these two types of tracts were significant. Overall, the H-POC tracts were 1.87 times more likely to have an additional food outlet than VL-POC tracts (IRR 1.870; 95% CI = 1.044, 3.348). The most significant findings resulted from the comparisons of the following types of food outlets: H-POC tracts were 17.32 times more likely than VL-POC tracts to have an additional limited-assortment store (IRR 17.320; 95% CI = 3.525, 85.100), 7.79 times more likely to have an additional mass merchandiser/supercenter (IRR 7.787; 95% CI = 2.219, 27.321), and 4.83 times more likely to have an additional traditional supermarket/large grocery store (IRR 4.830; 95% CI = 1.726, 13.515).

The H-POC tracts were also significantly more likely than the VL-POC tracts to have several categories of small food stores. Hence, H-POC tracts were 4.02 times more likely to have dollar and variety stores than VL-POC tracts (IRR 4.015; 95% CI = 1.734, 9.299). The H-POC tracts were 2.27 times more likely (IRR 2.266; 95% CI = 1.012, 5.074) to host an additional small grocery, convenience, or corner store and 2.16 times more likely (IRR 2.164; 95% CI = 1.177, 3.979) to have an additional gas station that sells food than VL-POC tracts.

Significant differences were also found between the H-POC and VL-POC tracts in the urban farming/community garden space and the emergency food assistance sector. We found that H-POC tracts were 4.81 times more likely to have an additional urban farm or community garden than VL-POC tracts (IRR 4.810; 95% CI = 1.733, 13.351). The H-POC tracts were 8.18 times more likely than the VL-POC tracts to have an additional food pantry/soup kitchen (IRR 8.177; 95% CI = 2.770, 24.136).

The restaurant and food service sector analysis revealed two instances where the H-POC tracts differed significantly from the VL-POC tracts. Hence, H-POC tracts were 2.25 times more likely to have an additional fast-food restaurant (IRR 2.249; 95% CI = 1.017, 4.975) and 1.93 times more likely to have an additional full-service restaurant than VL-POC tracts (IRR 1.934; 95% CI = 0.978, 3.826).

Only two significant findings were revealed in the VL-POC and VH-POC comparisons. The results show that VH-POC tracts were 71.6% less likely than VL-POC tracts to contain an additional food outlet (IRR 0.284; 95% CI = 0.096, 0.838). The VH-POC census tracts were also 85.2% less likely to have an additional full-service restaurant than the VL-POC tracts.

Although insignificant, the study found that L-POC tracts were 38% less likely and VH-POC tracts were 51% less likely to contain an extra liquor or party store than the VL-POC tracts. The study also found that L-POC, H-POC, and VH-POC tracts were less likely to have an additional takeout restaurant than VL-POC tracts.

We assessed the role of other independent variables—educational attainment, median household income, and population density—in predicting which type of census tract additional food outlets are likely to be (

Table 7. Incidence Rate Ratios for the Full Model Showing Census Tracts, Median Income, Educational Attainment, and Population Density.

Food Outlet Type a,b	Percent of Population with High School Education				Median Income per 1000 Dollars				Population Density per Square Kilometer
	Incidence Ratio Rate (IRR)	95% Confidence Interval (CI)		p-Value e	Incidence Ratio Rate (IRR)	95% Confidence Interval (CI)		p-Value	Incidence Ratio Rate (IRR)	95% Confidence Interval (CI)		p-Value
		Lower	Upper			Lower	Upper			Lower	Upper
d Total number of food outlets	1.003	0.981	1.027	0.778	0.988	0.981	0.996	0.003 *	1.000	1.000	1.000	0.007 **
c Traditional supermarkets and large groceries	1.033	0.963	1.108	0.362	0.990	0.971	1.010	0.346	1.000	1.000	1.000	0.227
c Limited-assortment stores	0.989	0.929	1.052	0.719	0.977	0.948	1.007	0.137	1.000	0.999	1.000	0.163
d Mass merchandisers and supercenters	1.333	1.090	1.630	0.005 **	0.992	0.960	1.025	0.644	1.000	0.999	1.000	0.235
c Gas stations with food	0.999	0.955	1.045	0.975	0.988	0.974	1.003	0.114	1.000	0.999	1.000	0.042 *
c Liquor stores and party stores	0.988	0.955	1.021	0.464	0.973	0.954	0.992	0.005 **	1.000	1.000	1.000	0.171
c Small groceries, convenience, and corner stores	0.982	0.951	1.013	0.249	0.989	0.973	1.006	0.206	1.000	1.000	1.000	0.119
d Pharmacies or drug stores	0.981	0.942	1.022	0.358	0.989	0.973	1.007	0.226	1.000	1.000	1.000	0.153
c Dollar stores and variety stores	0.990	0.937	1.046	0.718	0.991	0.973	1.010	0.366	1.000	0.999	1.000	0.070
c Bakeries	1.097	0.988	1.218	0.084	0.993	0.974	1.013	0.519	1.000	1.000	1.000	0.819
c Ice cream parlors	0.950	0.885	1.020	0.158	1.006	0.983	1.030	0.604	1.000	0.999	1.000	0.397
d Full-service restaurants	1.029	0.983	1.077	0.217	0.988	0.976	1.000	0.060	1.000	1.000	1.000	0.134
d Fast-food restaurants	1.012	0.970	1.057	0.569	0.993	0.981	1.006	0.307	1.000	1.000	1.000	0.181
d Takeout	1.002	0.953	1.053	0.937	0.976	0.959	0.994	0.009 **	1.000	1.000	1.000	0.093
d Coffee, tea, and juice shops	0.995	0.960	1.031	0.783	0.993	0.978	1.007	0.317	1.000	1.000	1.000	0.716
d Banquet halls and hotels	1.075	0.983	1.175	0.114	0.979	0.959	1.000	0.049 *	1.000	0.999	1.000	0.047 *
d Bars and clubs	1.035	0.949	1.129	0.438	0.975	0.954	0.996	0.022 *	1.000	1.000	1.000	0.206
d Urban farms and community gardens	0.973	0.931	1.017	0.218	0.974	0.952	0.997	0.025 *	1.000	1.000	1.000	0.092
c Farmers’ markets and produce markets	0.987	0.929	1.049	0.672	0.996	0.973	1.019	0.725	1.000	1.000	1.000	0.644
c Food pantries or soup kitchens	0.990	0.957	1.025	0.568	0.992	0.974	1.010	0.387	1.000	1.000	1.000	0.205
c Food banks/distribution	1.003	0.925	1.087	0.947	0.988	0.956	1.020	0.453	1.000	1.000	1.000	0.940
c School cafeterias	1.040	0.973	1.112	0.251	0.987	0.968	1.006	0.184	1.000	1.000	1.000	0.227
c Retirement communities and homes	1.083	0.956	1.227	0.210	0.985	0.961	1.010	0.235	1.000	0.999	1.000	0.168
d Childcare	0.985	0.939	1.033	0.526	0.994	0.975	1.015	0.582	1.000	1.000	1.000	0.362
d Fitness centers, gyms, and health centers	1.276	1.018	1.601	0.035*	0.954	0.920	0.989	0.010 **	0.999	0.998	1.000	0.060
c Wholesalers	1.062	0.965	1.168	0.217	0.963	0.930	0.997	0.034 *	0.999	0.998	1.000	0.023 *
d Manufacturers, processors, and distributors	0.994	0.898	1.100	0.907	0.980	0.946	1.014	0.249	1.000	0.999	1.000	0.200

Notes: ^a Each row represents a separate model that adjusted for population density, total number of households, median income, and the percent of the population over 25 that has at least a high school education. ^b Model for the following food outlets childcare, manufacturers, processors and distributors, farmers’ markets and produce markets, food bank/distribution, and ice cream parlors were not significant; p-values reported as needed. ^c Indicates models that followed a Poisson regression. ^d Indicates models that followed a non-binomial regression. ^e Significance levels: * p-values < than α = 0.05, ** p-values < than α = 0.01.

). Educational attainment was a significant predictor for only two outlet types. For every one-point increase in the percentage of people who obtained at least a high school education, L-POC tracts, there was a 33.3% increase in the likelihood of having an additional mass merchandiser or supercenter (IRR 1.333; 95% CI = 1.090, 1.630). With each one-point increase in the percentage of people who obtained a high school education, there was a 27.6% increase in the likelihood of having an additional fitness center/gym/health center (IRR 1.276; 95% CI = 1.018, 1.601).

The analysis revealed an inverse relationship between median income and the distribution of eight food outlet types. The results show that for every USD 1000 increase in median income, there was a 1.2% reduced likelihood (IRR 0.988; 95% CI = 0.981, 0.996) of having an extra food outlet.

For each USD 1000 increase in median income, there was also a 2.7% less likelihood (IRR 0.973; 95% CI = 0.959, 0.994) of containing an extra liquor or party store and a 2.6% less chance of having an additional urban farm or community garden (IRR 0.974; 95% CI = 0.952, 0.997). There is also a 2.4% less likelihood (IRR 0.976; 95% CI = 0.959, 0.994) of containing a takeout restaurant. A USD 1000 increase in median income was also associated with a 2.1% decrease in the likelihood of having an additional banquet hall or hotel (IRR 0.979; 95% CI = 0.959, 1.000) and a 2.5% decrease in the likelihood of having an additional bar or club (IRR 0.975; 95% CI = 0.954, 0.996). As income increases, the lower the likelihood of having additional fitness centers/gyms/health centers (IRR 0.954; 95% CI = 0.920, 0.989) and wholesalers (IRR 0.963; 95% CI = 0.930, 0.997).

There were four instances where population density had significant effects on the distribution of food outlets. For each one-point increase in the population density, there is a 1% decrease in the likelihood of having an additional food outlet. There is a similar likelihood of a decrease in additional gas stations selling food, having an additional banquet hall or hotel, or having an additional wholesaler.

5. Discussion and Implications

5.1. The Racial Characteristics of Census Tracts and the Distribution of Food Outlets

This article analyzed the food environment of Lansing, East Lansing, and the suburbs. It identified 53 sub-categories of food outlets in the study area and analyzed their association with census tract demographic characteristics. The study examined 1647 food outlets; 579 were in Lansing, 220 in East Lansing, and the remainder in the surrounding townships.

No previous study of food access in the Lansing metropolitan area has examined such an expansive list of food outlets. Instead, prior research focused on residents’ food purchases and nutrition intake [128], classified retailers based on food availability [167], analyzed geospatial techniques used in food studies [124], and examined residents’ perception of access to fresh and frozen food [168]. Our work expands the scope of these earlier studies by performing extensive documentation and mapping of food retailers to produce a more comprehensive picture of the food landscape in the metropolitan area.

We found a robust assortment of food outlets in the study area. There were roughly 4.18 food outlets per 1000 residents. However, the analysis shows that access is not evenly spread. For instance, the VH-POC census tracts had 1.57 food outlets per person. The data show that VH-POC residents, more than 60% of whom are People of Color, have significantly reduced access to food outlets than residents in census tracts with lower percentages of POC. The VH-POC census tracts had 23 food outlets; the tracts were devoid of several major food categories and outlet types. These findings lend some support to other studies that find that predominantly POC communities lack access to food outlets that sell healthy and affordable foods [25,93,94,95,128,169,170,171,172,173]. However, we should exercise caution in interpreting the results. Our data show that the percentage of POC in the census tracts matters. We found that POC residing in VL-POC, L-POC, and H-POC census tracts have access to a broader range of food outlets than those in the VH-POC tracts.

The H-POC census tracts are intriguing and vital, as they represent parts of the cities where between 40.01% and 60% of the residents are People of Color. However, the tracts have the largest number of significant findings for having a greater density of several types of food stores or being likely to have additional food outlets than other tracts. The findings suggest that future studies should examine more gradations in racial characteristics, income, or education levels to better understand which residents have the least access to food.

Research on travel distance to purchase food shows that food shoppers do not always buy food from the retailers nearest their residence and often travel over three miles to make food purchases [25,31,32,128,151]. Findings like these led us to examine food outlets within the study cities and eight kilometers (about five miles) beyond their boundaries. Extending the boundaries of our study area helped us capture the full range of food outlets where residents of the metropolitan area are likely to shop for food. We found that many large food retailers, such as mass merchandisers and supercenters, are located in the townships, but close to the city limits. Because the food retailers build their establishments along the highways and other major thoroughfares, residents with access to transportation can reach these quickly.

Despite the heavy focus on supermarkets and large grocery stores in past food access studies [18,20,25,26,27,92,93,94,95,96,97,174] and the importance that area residents place on them [128], we found that supermarkets and large grocery stores constituted a minor component of the food outlets. This food outlet category accounted for 5.6% of the food environment of the entire study area. Our study corroborates the findings of Taylor & Ard [17], Taylor et al. [40], and Short et al. [112].

There are lingering impacts of redlining in Lansing. Despite integrating many of the city’s neighborhoods, some formerly redlined communities still have high concentrations of poor People of Color [70]. Currently, Blacks and Hispanics live in such neighborhoods. The low Black–Hispanic D score of 16.4 in Lansing attests to the fact that these two ethnic groups tend to live in the same census tracts. Our results indicate that formerly redlined neighborhoods in which People of Color currently comprise more than 60% of the population have attenuated access to many kinds of food outlets.

5.2. Moving beyond the Focus on Supermarkets and Large Grocery Stores

We found 36 traditional supermarkets and grocery stores in 31 census tracts; the 26 limited-assortment food stores were in 19 census tracts, and 25 mass merchandisers and supercenters were in 16 census tracts. The category—supermarkets and large grocery stores—constituted 7.6% of Lansing’s and 3.6% of East Lansing’s food outlets. Lansing seems to be more endowed with this category of food outlets than other Michigan cities that have been assessed. For instance, a study of Detroit’s food outlets found that supermarkets and large grocery stores comprised 2.7% of the city’s food venues [17]. In Flint, this category of food retailers comprised 2.2% of the food outlets studied [40].

Our study found differential access to supermarkets and large grocery stores that indicate some underserved areas. We found no traditional supermarkets or large grocery stores in the five VH-POC census tracts. These census tracts were also devoid of mass merchandisers or supercenters; there was only one limited-assortment food store in the tracts. This means that the category of supermarkets and large grocery stores is mainly absent from the census tracts with the highest percentages of POC in the study area.

Lansing–East Lansing mirrors the findings of similar studies in Flint and Detroit. In Flint, in census tracts where Blacks comprised more than 40% of the population, residents had lower access to supermarkets and large grocery stores [40]. Similarly, in Detroit, the percentage of Black residents in a neighborhood was associated with the ratio of residents to supermarkets and large grocery stores. Hence, neighborhoods wherein more than 40% of the residents were Black had lower access to these types of food outlets [17].

Ergo, other studies make claims about supermarkets and large grocery store access that some of our data support. Earlier research suggests that poor, urban, Black, and Hispanic communities have less access to supermarkets and large grocery stores than high-income White communities [20,25,27,92,93,94,95,96,97,174,175]. Other researchers urge us to think more broadly about how food access intersects with health and wellbeing [176].

Our study area’s food landscape illustrates the shortcomings of relying solely or almost exclusively on supermarkets and large grocery stores to define food access. To assume that all the census tracts that do not have supermarkets and large grocery stores are food deserts would be a faulty conclusion to draw. Table 2 above shows that a complex food landscape can allow people to obtain a variety of foods, even if there are no supermarkets or large grocery stores in their census tracts. Hence, we contend that food insecurity studies focusing exclusively on supermarkets and large grocery stores may overestimate these outlets’ contributions to urban dwellers’ food access. Focusing exclusively on supermarkets and grocery stores overlooks key aspects of urban food landscapes that should be accounted for.

5.3. Restaurants Dominating the Food Environment

Our study found that restaurants are the most dominant food outlets in the study area; they constituted 38.7% of Lansing’s, 48.6% of East Lansing’s, and 52% of the food landscape of the peri-urban area. Nevertheless, researchers such as Duvall et al. [167] and Goldsberry et al. [124] did not include restaurants in their study of Lansing’s food landscape. Restaurants also dominate the food landscape in Detroit. Taylor & Ard [17] found that restaurants constituted 35.6% of the Detroit food outlets studied. Restaurants also comprised 26.5% of Flint’s food outlets [40]. Rybarczyk et al. [31] also found restaurants crucial to ethnic food outlets in Grand Rapids and Flint.

Several scholars argue that there are more fast-food restaurants in Black and poor neighborhoods than in White and high-income neighborhoods [94,96,177,178,179,180,181]. However, our results are more nuanced than the findings of earlier studies of race and the distribution of food outlets suggest. For instance, we found a nonlinear relationship between the number of fast-food restaurants and the census tract’s racial composition. There were 1.08 fast-food outlets per 1000 residents in VL-POC census tracts. It decreased to 0.93 fast-food outlets per 1000 persons in L-POC tracts. The highest ratio was found in the H-POC census tracts—there were 1.47 fast-food outlets per 1000 residents. The VH-POC tracts had no fast-food outlets. In other words, the census tracts with the highest percentage of POC had no fast-food outlets. This is noteworthy, as the VH-POC census tracts also had the lowest household income and highest poverty rate of all assessed tracts.

5.4. Defying the Food Swamp and Food Oases Predictions

Some researchers go as far as to suggest that fast-food restaurants, mini marts, and gas stations selling food create food swamps that reduce the likelihood of supermarkets and grocery stores operating in Communities of Color [104,105,106,107,108,109,111,127].

Our findings run contrary to the food swamp thesis. We found no evidence of food swamps in the census tracts with the highest percentage of POC in our study area. There was also no evidence of food oases [112,155] in such tracts. The VH-POC census tracts had no small grocery stores, convenience stores, corner stores, or mini marts. These tracts had the lowest number of liquor stores or party stores. Of the four groups of census tracts studied, the VH-POC tracts had the lowest number of gas stations that sell food.

If hyper-concentrated clusters of food retailers form junk food strips or food swamps in our study areas, they are more likely to occur along major thoroughfares in the L-POC and H-POC census tracts. Concomitantly, if there are food oases, they are also most likely to occur in the L-POC and H-POC tracts. It is beyond the scope of this paper to conduct these analyses. However, future research should explore more fully the racial composition of census tracts or neighborhoods and the clustering of food outlets.

5.5. Alternative Food Sources

Our study corroborates Reed’s [128] finding that there was a robust infrastructure of alternative food venues in Lansing. Forty-eight percent of the participants in Reed’s study shopped at farmers’ markets, while thirty-six percent purchased items from health food stores. Our study found that farms, community gardens, farmers’ markets, and produce vendors comprise 5.5% of the area’s food environment. Consequently, this vibrant segment of the food landscape that sells or gives away fresh and healthy foods should not be disregarded. Taylor et al.’s [85] study of community gardens illustrates the vital role these food sources serve in providing free food to Michigan’s residents. These food venues report increased demand for free food during the pandemic. See Braswell [182], Giraud et al. [183], Lal [184], and Niles et al. [185] for studies and discussions of community and home gardens and urban agriculture.

A study of farmers [186] and farmers’ markets [187] in the state also reveals that farms and farmers’ markets are also crucial food sources for the state’s residents. The significance of these came into sharp focus during the pandemic. At the height of the pandemic, Michigan’s farmers and farmers’ markets were asked to assemble and deliver free boxes of locally grown farm produce to needy residents. The federal government sponsored the Farmers to Families Food Box Program (FFFBP), which paid for and helped to facilitate food distribution. Supply chain food entities played critical roles in the execution of the FFFBP. Hence, food aggregators, like Eastern Market in Detroit, collected, processed, and packaged the food boxes. Emergency food assistance organizations also helped, as food pantries and food banks were pressed into service to help deliver the food boxes. In addition, pop-up structures and mobile food trucks were used to distribute food boxes [44,186,188].

5.6. Where There Are No Food Outlets

According to the United States Department of Agriculture, parts of Lansing are food deserts [113,128], however, the above findings do not support this claim. The study found 11 census tracts with no food outlets; all but 1 of these were in suburban tracts. One was in Lansing; none were in East Lansing. We describe the tracts containing no food outlets as extremely low-access tracts. The configuration of the suburban ELA census tracts is reminiscent of the original description of a food desert that Cummins and McIntyre [189] popularized in the United Kingdom. They used the concept to characterize suburban housing developments bereft of churches, stores, or community centers. As mentioned earlier, we refrain from describing neighborhoods we study as food deserts. Although we found no food outlets in the ELA tracts, it could be premature to assume that no food sources exist in these census tracts.

5.7. Multivariate Analyses

We assessed the likelihood of finding additional food outlets based on the racial composition of census tracts. Our IRR findings suggest greater nuances than have been reported elsewhere. First, our study found that census tracts containing 40.1–60% POC (H-POC tracts) were 1.87 times more likely to contain an additional food outlet than census tracts with 0–20% POC (VL-POC tracts). Second, the H-POC tracts were significantly more likely to contain several food outlet types than the VL-POC tracts. However, our study reveals that the census tracts with the highest percentage of POC—the VH-POC tracts that contain 60.1% or more POC—are 0.72 times less likely to have an extra food outlet than the VL-POC tracts.

Our findings differ from those of LaVeist and Wallace [190], who report that liquor stores were disproportionately located in Black census tracts. Our study found that the VH-POC tracts are 0.51 times less likely to contain an extra liquor or party store than the VL-POC census tracts. In fact, in the study area, the VL-POC census tracts are more likely to have liquor and party stores than the three other categories of tracts they are compared to.

We also examined the contribution of median income and educational attainment to the distribution of food outlets. The study found that the likelihood of having an additional food outlet tended to decrease as income increased. Our findings provide support to studies that report that low-income neighborhoods have less access to food outlets than high-income neighborhoods [104,105,106,107,108,109,110,111]. Our study found that, generally speaking, the likelihood of having additional food outlets in census tracts was not strongly associated with educational attainment.

5.8. Store Closures

We collected data on Lansing, East Lansing, and the peri-urban area during the COVID-19 pandemic. This led us to examine food store closures. The study found 257 closed food outlets. Of these, 137 were closed in Lansing, 61 in East Lansing, and 59 in the surrounding townships. The extensive closure in Lansing is concerning. The extent of the closures in the two cities and peri-urban areas mirrors what Bell and Taylor [29] found in their Flint study.

5.9. Local Initiatives

Food activists know that small grocery stores, corner stores, and mini marts are more common in some neighborhoods than traditional supermarkets and grocery stores. Consequently, they are trying to develop small-scale food operations that stock and sell fresh, locally grown, affordable, healthy foods. Michigan’s two largest cities—Detroit and Grand Rapids—have successful healthy corner initiatives [23,191]. Lansing and East Lansing have not yet launched a healthy corner store initiative, but have neighborhood food markets and food cooperatives that provide access to healthy foods. Examples are East Lansing’s Campbell’s Market Basket, Fresh Thyme, and an Asian grocery store—Fresh International Market. Our findings suggest that mid-sized and small cities like Lansing and East Lansing should also develop healthy corner store programs.

Lansing and East Lansing have turned to farmers’ markets to help increase access to healthy and affordable foods for their residents. Farmers’ market vendors are accepting the Supplemental Nutrition Assistance Program (SNAP) and WIC EBT cards (or Bridge Card, as it is commonly called in Michigan) and are participating in the Double-Up Food Bucks (DUFB) program. Customers who are eligible for DUFB receive tokens or coupons worth USD 20.00. For each USD 1 spent at the farmers’ market, DUFB customers can get an additional USD 1 worth of food, up to a maximum of USD 20.00 [192,193].

For example, the farmers’ market operated by the Allen Neighborhood Center is located in a low-income, POC, immigrant community on Lansing’s Eastside. The center serves this target audience with robust programming to increase food access that focuses on the farmers’ market, community-supported agriculture (CSA) and the veggie box, urban farming, community gardening, and an incubator kitchen [194]. The center is guided by food sovereignty and justice principles [17,24,116,117,122]. Three other farmers’ markets (Farmers’ Market at the Capitol, South Lansing Farmers’ Market, and the East Lansing Farmers’ Market also participate in DUFB [54].

In addition to the farmers’ markets, one grocery store participates in the DUFB program. Great Giant in Lansing allows customers who purchase fruits and vegetables with their SNAP EBT card to get USD 1 for each USD 1 spent on these produce (up to USD 10 per day) on a Double Up Card. Several grocery stores in Detroit and the Tri-County Area flanking that city (Wayne, Macomb, and Oakland counties) already participate in the DUFB program [54]. Such a program can be expanded to include more supermarkets and large and small grocery stores.

Our study reveals that restaurants dominate the Lansing–East Lansing metropolitan area’s food landscape. This implies that restaurants should be incorporated more fully into efforts to improve food access. Lansing is one of 12 cities in Michigan where restaurants participate in the Restaurant Meal Program (RMP). The RMP’s Food Assistance Program allows residents who are elderly, have disabilities, or are homeless to be able to purchase restaurant meals with their EBT cards. Two of the twenty-five restaurants in the state participating in this program are in Lansing. The two participating Lansing restaurants are Just Munchies and Siraj Cuisine [195]. There is potential to expand this program. However, since some restaurants participating in the RMP are fast-food establishments, like McDonald’s, care should be taken to ensure fresh and healthy food options are available for EBT customers to purchase.

Lansing also has the NorthWest Initiative, which partners with the Michigan State College of Human Medicine to provide more than 900 boxes of produce to residents of the metropolitan area. The initiative also distributes bread and operates the Lansing Mobile Farmers’ Market [196].

Increasingly, hospitals and medical centers are providing patients with free or reduced-price healthy foods through programs aimed at improving nutrition and reducing food insecurity. Since 2021, Lansing’s Sparrow Hospital has developed a program to identify and provide food staples to food-insecure pregnant and new mothers and their families. Program participants receive groceries, a cookbook, a crockpot, and healthy eating messages [197].

5.10. Strengths and Limitations

Our study has several strengths. We analyzed an extensive list of food outlets and food categories. This approach provides a more accurate depiction of the Lansing metropolitan area’s food landscape than we have had before. The study contributes to the food access literature because it focuses on a type of metropolitan area often overlooked in food access studies. Lansing/East Lansing is mid-sized, with intact inner-city neighborhoods, and is not comprised primarily of poor residents or POC. By examining cities like this, we demonstrated that local food landscapes are more complex and nuanced than some early food access studies imply.

Nonetheless, the study has some limitations. We acknowledged earlier that we did not study all the possible food sources; consequently, residents might have access to food derived from sources not included in our study. We studied the distribution of a wide range of food outlets, but did not examine the content in the venues. Hence, we did not assess which carried healthy and affordable foods and which did not.

The data were collected during the COVID-19 pandemic. The pandemic triggered many store closures; these may have impacted the results. However, pandemic-related food store closures are not unique to the Lansing–East Lansing area. As Bell and Taylor [29] report, store closures are being documented in other parts of Michigan and around the country. Lastly, we follow the techniques outlined in Rosencrants et al. [160], Taylor et al. [40], and Bell and Taylor [29]. However, we note that using proportional allocation to estimate the population size of census tract fragments may cause either over- or underestimation.

5.11. Future Directions

Our findings suggest that more studies that examine the interaction effects among race and income and access to food should be conducted. Although we found a limited association with educational attainment, this variable could be of greater significance in study areas with lower educational attainment. We will conduct comparisons of college towns in future papers, as we collected data in several Michigan and New England college towns.

We will examine the extent to which food retailers have closed or opened new food outlets or re-opened new ones in Lansing and East Lansing. This will help us to understand the extent of vanishing or emergent food infrastructures. We also plan to analyze the distribution of ethnic food outlets to determine where they are located and how they impact food access. We will also analyze other variables, such as food outlet size (in square feet), sales volume of the food outlet, number of employees at the food outlet, ethnicity of the owner of the food outlet, gender of the food outlet owner, and cuisine type of restaurants and other food service operations, to determine how they are distributed in the study area.

Studies like ours could be complemented with transportation analysis examining the relationship between public transportation (bus routes and bus stops) and the location of various food outlets. Investigations that test the food swamp or food oasis theses could contribute significantly to this genre of research.

6. Conclusions and Recommendations

Food access researchers have largely ignored Lansing and East Lansing, focusing instead on more well-known cities. Our study shows that it is important to look at places like the Lansing metropolitan area that are not areas of extremes. The cities are not characterized by extreme poverty, deindustrialization, or racial imbalance and are not oversized. By examining two such cities, we cast doubt on the assumption that areas with robust numbers of POC are usually destitute and devoid of healthy food retailers.

Although the USDA describes parts of the Lansing metropolitan area as food deserts, our study shows that the food desert designation is misleading. It shows the importance of examining more than supermarkets and large grocery stores when describing the food landscape. The study reveals a diverse food landscape in the study cities. Although the racial composition of the census tracts is an important factor in food access, the relationship is not linear. In contrast, the relationship between income and food access is more straightforward and linear.

We recommend that future investigations extend the study area beyond the city limits. More studies of transportation access and commuting between the city and suburbs should be conducted to determine how these dynamics influence access to food and purchasing behaviors. We also strongly suggest that food access researchers focus on a broad range of food sources and examine multiple independent variables to help further our understanding of urban food environments more completely. More research on the food content in various food retailers would also help assess food access. Greater efforts should also be made to expand healthy corner store initiatives, and more studies should be conducted to determine if and how these undertakings increase access to healthy and affordable food.