Community Assets and Relative Rurality Index A Multi-Dimensional Measure

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Journal of Rural Studies 97 (2023) 322–333

Contents lists available at ScienceDirect

Journal of Rural Studies


journal homepage: www.elsevier.com/locate/jrurstud

Community assets and relative rurality index: A multi-dimensional measure


of rurality
Katherine S. Nelson a, *, 1, Tuan D. Nguyen b, 1
a
Department of Geography & Geospatial Sciences, Kansas State University, 1002 Seaton Hall, Kansas State University, 920 N. Martin Luther King Jr. Drive, Manhattan,
KS, 66506-2904, USA
b
College of Education, Kansas State University, 261 Bluemont Hall, 1114 Mid-Campus Drive, Manhattan, KS, 66506, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Rurality is often viewed as presenting challenges to community sustainability, well-being, and equity. To address
Rurality the unique challenges of rural areas, policies and practices have been designed specifically for application in
Measure of rurality places designated as “rural”. Yet what is “rural”? Some recent measures of rurality have gone beyond a
Services
dichotomous rural-urban divide conceptualization of rural communities. However, most measures still empha­
Amenities
Block group
size proximity to metropolitan areas and population density as the primary components of rurality. Few studies
consider the critical role that services and amenities play in the life of a community. We suggest a new measure
based on the concept of sustainable development that integrates measures of environmental, social, and eco­
nomic resources. We present the Community Assets and Relative Rurality index for census block-groups in the
coterminous United States and illustrate how this measure is consistent with existing measures of rurality yet
offers additional insight into issues of sustainable rural development.

The measure of rurality is important for researchers and policy­ aware of rural development policies that take this issue into account (Li
makers as it influences the identification of at-need populations and et al., 2015). In the healthcare industry, there are many concerns with
study populations; how outcomes of policies, environmental changes, respect to the qualifications and quantity of healthcare workers, access
and social disturbance are evaluated; and the results of these evalua­ to healthcare facilities, and the health of people in rural communities
tions. Consequently, measures of rurality influence the distribution of (Higgs, 1999; Inagami et al., 2016; Kulig et al., 2008; Mao et al., 2015;
funding and other aid across nations and states. From a policy stand­ Zhao et al., 2019). During the COVID-19 pandemic, research finds the
point, the concerns associated with rurality are typically related to issues effects of the pandemic on rural life are severe, with significant negative
of community sustainability, well-being, equity and access (Brown and effects on mental and physical health, access to healthcare, life satis­
Cromartie, 2017; Minore et al., 2008). From an environmental sustain­ faction, economic outlook, and unemployment (Henning-Smith, 2020;
ability approach, rural areas are associated with issues of land and water Mueller et al., 2021). While these issues are not restricted to rural areas,
resource exploitation and subsequent human and environmental health prior work suggests rural areas tend to experience these issues more
outcomes (Brown et al., 2005; Chen et al., 2018). In education, rural severely than urban areas (Berenguer et al., 2005; Ellis and Biggs, 2001).
education is associated with concerns regarding access to schooling, the Moreover, as rural places are the backbone of the national economy,
recruitment and retention of teachers, and financial support of schools providing the raw materials and resources that fuel trade and industry,
(Crouch and Nguyen, 2021; Nguyen, 2020; Sher, 2019). In economic and as they represent the lion’s share of the land area in the U.S. (Brown
development, job growth, retention and diversification, labor force et al., 2005; Henderson, 2002), how rurality is defined and measured
availability and skills, as well as tax base and incomes are concerns matters greatly for policy and practice and consequently for the people
frequently associated with rural areas (Ellis and Biggs, 2001; Galluzzo, who live in these communities and the rest of society.
2018). In particular, there have long been concerns that highly rural Therefore, we propose a new measure that considers the underlying
counties in the U.S. may be marginalized, especially in economic physical infrastructure of communities that directly reflects (dis)ad­
development, resulting in recognition that policymakers need to be vantages in the provisioning of services and amenities (henceforth,

* Corresponding author.
E-mail address: ksnelson@ksu.edu (K.S. Nelson).
1
Equal contribution.

https://doi.org/10.1016/j.jrurstud.2022.12.025
Received 13 July 2022; Received in revised form 18 November 2022; Accepted 12 December 2022
Available online 21 December 2022
0743-0167/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
K.S. Nelson and T.D. Nguyen Journal of Rural Studies 97 (2023) 322–333

assets). We suggest that provisioning of assets is a determinant of the area, the population density, distance from metropolitan areas, and
community (dis)advantages and therefore provides a way of examining the built-up percent of the area (Waldorf, 2006; Waldorf and Kim,
current conditions in these communities and also improves under­ 2015). However, these measures still rely heavily on remoteness and
standing of the future potential for these communities to reduce their population density as the primary components of rurality.
disadvantages by leveraging existing infrastructure. Drawing from the Many have argued that, in post-industrial societies, rurality must be
concept of sustainable development, the proposed measure integrates multi-dimensional to include demographic, economic, institutional, and
natural resources, social, and economic infrastructure resources of local sociocultural measures (Brown and Cromartie, 2002; Lichter and Brown,
areas with the potential to contribute to inclusive well-being (Matson 2011; Minore et al., 2008; Mountrakis et al., 2005). Rurality as a concept
et al., 2016). This new index of rurality called Community Assets and is more than just how physically isolated people are from others. For
Relative Rurality (CARR) consists of the primary components of previ­ instance, Brown and Cromartie (2017) argue the natural environment is
ous indices and additional measures of assets available in an area. The a component of rurality as well as the size of the local economy, the
measure merges traditional notions of rurality with the idea of infra­ autonomy of the local economy, the public sector capacity, the beliefs
structure development deficits, characterizing variation in “rural” and values of the local population, and the diversity of the population.
spaces and “urban” spaces according to population and remoteness as Others argue existing measures may mask significant heterogeneity and
well as the presence of community assets that support provisioning of problems within rural populations, particularly for low-income resi­
resources key to supporting well-being and community persistence. We dents, minority populations, and seniors (Hall et al., 2006; Minore et al.,
develop this continuous measure for the coterminous U.S. at a high, 2008). Relatedly, Kulig et al. (2008) find community characteristics, and
block-group, spatial resolution. health and human resources are also key components of rurality and
The rest of the paper is organized as follows. In the next section, we remoteness, but that not everything associated with rural living is
briefly provide the background on how measures of rurality have been considered negative or disadvantaged. In their study, they find the idea
conceptualized and the importance of rural sustainability with regards of “country living” and more “personal” interpersonal interactions are
to measures of rurality. Then we discuss the overaching framework for more associated with rural settings and are considered positive char­
the proposed CARR index and discuss its advantages and limitations. acteristics that separate rural from urban settings.
After that, we provide the basic descriptive data about the CARR index
and discuss the spatial pattern of CARR across the U.S. We select and 1.2. Rural sustainability and measures of rurality
discuss several cases that contrast the differences between the CARR
index and current measures of rurality. Lastly, we discuss our results, the Within the literature on rural development, it has been acknowl­
implications for policy and practice, and directions of future research. edged that service infrastructure is critical for development as it allows
for expansion and maintenance of population and the growth of the
1. Literature review economic base of a community (Nirandjan et al., 2022). Ashley and
Maxwell (2001) note rural spaces tend to not only be associated with
1.1. Conceptualizing measures of rurality long distances and high transaction costs but also with poor infrastruc­
ture. Investment in transportation, communication, education, and
For nearly a century, researchers have been trying to define rurality health infrastructure and development of recreational amenities have
with more precision and practical use, and consequently, rurality has also been suggested as strategies to support rural non-farm economies.
been numerously defined (Mountrakis et al., 2005; Nelson et al., 2021; Relatedly, infrastructure providing benefits to under-served commu­
Woods and Heley, 2017).2 Originally, rurality was used as a way to nities, particularly infrastructure that supports essential services like
define a space, physical and otherwise, clearly delineated from energy, water, and transportation services, is also seen as a key aspect of
urban-industrial areas (Minore et al., 2008). For practical purposes, sustainable and inclusive development (Nirandjan et al., 2022; Ramos
most measures of rurality define the rural-urban divide based on pop­ and Uitermark, 2021; Siddiqi and Collins, 2017). While increased pop­
ulation density and size, in part because population size and density ulation and income also drive further development, we suggest the
information are readily available and easy to apply (Brown and Cro­ presence of some initial infrastructure development is a prerequisite for
martie, 2017; Minore et al., 2008; United Nations, 1999). For instance, achieving a stable community with high levels of well-being. This initial
the most basic definition of a rural area is a non-urban area, where urban development of physical infrastructure provides a level of
is based on population density in an enclosed space (Minore et al., semi-permanence that can “lock-in” sociotechnical systems with the
2008). Similarly, the U.S. Office of Management and Budget (OMB) potential to benefit communities’ well-being and persistence. While the
defines rurality in terms of metropolitan areas as the counties or clusters term “sociotechnical lock-in” often has a negative connotation, and it is
of counties containing at least one urban area with a population of at certainly true that lock-in of highly specialized systems can hurt com­
least 50,000 (U.S. Dept. Of Health & Human Services, 2022). munities in the long run (Cairns, 2014; Meynard et al., 2018), we suggest
Recognizing the limitations of population variables to represent that a moderate degree of lock-in (particularly of basic or generalizable
diverse challenges of rural life, measures of rurality emerging in recent services) can have a positive effect on small, rural communities by
years have attempted to go beyond a dichotomous rural/urban divide or enhancing stability.
threshold-based conception of rural communities. Measures that expand While others argue services and amenities should be included as part
on a simple population-based rural-urban dichotomy, such as the Rural- of the measure of rurality (Caschili et al., 2015; Mountrakis et al., 2005;
Urban Continuum Codes (RUCC) and the Urban Influence Codes (UIC), Scottish Executive, 2006), few measures of rurality have explicitly
have considered total urban population, size of largest city, and included service and amenity factors (Nelson et al., 2021). One notable
remoteness in terms of adjacency to metro areas (Minore et al., 2008; exception is recent work by the Urban Institute (Gold et al., 2021) that
Wunderlich and Rapporteur, 2016). Relatedly, other measures, such as uses information on infrastructure, educational and cultural facilities,
the Rural-Urban Commuting Area Codes (RUCA) and the Frontier and healthcare facilities, and resource extraction sites with demographic
Remote Codes (FAR), have also included commuting time (Wunderlich variables to develop a typology of rural census tracts in the U.S. Given
and Rapporteur, 2016). One of the most recent proposed indices of the critical roles services and amenities play in the life of a community,
rurality is the Relative Rurality Index (RRI), which is based on the size of and the disadvantages inherent to areas with little to no access to these
assets (Agarwal et al., 2009; Woods and Heley, 2017), additional work
considering these dimensions is needed. While the work by the Urban
2
Many other works have detailed the construction and development of Institute is a positive step, additional considerations regarding spatial
measures of rurality (see Cloke, 2006; Cloke et al., 2006). resolution, continuous approaches, and services and amenities should be

323
K.S. Nelson and T.D. Nguyen Journal of Rural Studies 97 (2023) 322–333

explored. availability, or migration (Kulig et al., 2008; Minore et al., 2008;


The spatial resolution of measures of rurality has been a frequently Mountrakis et al., 2005; Roscigno and Crowle, 2001). For instance, the
noted challenge for most commonly used measures (Nelson et al., 2021; Scottish Index of Multiple Deprivation (Scottish Executive, 2006) con­
Uhl et al., 2022). Most measures employ relatively course spatial units siders limited access to essential services as part of the “deprivation” of
such as counties (e.g. RUCC, RRI). This spatial coarseness can mask rural communities, and this index provides a measure of the extent to
heterogeneity within counties providing rurality rankings that are not which rural communities have better or worse access to services. In
reflective of all communities in a county. Recognizing this challenge, particular, in terms of healthcare, rural communities face several critical
higher spatial resolution measures at the census tract scale in the U.S. issues such as geographical remoteness, topographical features hinder­
(Doogan et al., 2018) or census meshblocks in New Zealand (Zhao et al., ing physical access to healthcare, and difficulties in recruiting and
2019) have been developed. However, in the U.S. census tracts can still retaining healthcare workers due to limited access to amenities (NRHA,
mask significant variation across space, particularly in rural areas where 2015; NRHA, 2017; Baker et al., 2011). Moreover, people also perceive
tracts tend to have larger areas due to their small population size. limited access to services and amenities to be associated with rural
communities, and costs in services tend to be higher in rural areas,
1.3. Framework for a community assets-based rurality measure further adding to their challenges (Kulig et al., 2008). In sum, services
and amenities are critical to rural communities, and the proposed CARR
The proposed CARR index is more comprehensive than previous index builds upon existing measures of rurality by including them as part
indices as it comprises multiple measures of three key categories that of its construct. While we are not the first to suggest that services and
arguably are central to rurality and how people experience it. These amenities be considered in measures of rurality, we are the first to pull
three categories are services, amenities, and relative rurality. Services together a rich and comprehensive set of open data sources of services
consist of transportation infrastructure, agriculture and natural re­ and amenities at a granular scale for the construction of a measure of
sources processing facilities, energy infrastructure, data and cellular rurality.
infrastructure, schools, banks, health services, supermarkets, and the While we calculate CARR for all block-groups in the U.S., we focus
like. Amenities include outdoor activities, entertainment, restaurants, our analysis of CARR on only block-groups with non-zero population in
gyms and sports centers, places of worship, and shopping centers. the 2019 U.S. American Community Survey as obtained from the Na­
Relative rurality is measured by population size and density, land use tional Historical Geographic Information System (NHGIS) database
development, and remoteness. Specific components of each category, (Manson et al., 2019). We first describe the range and spatial patterning
along with measurement scales, are detailed in the Supporting Infor­ of CARR, including descriptions of within state and within county het­
mation in Table S1. erogeneity. We then examine how CARR compares to common U.S.
Rurality may also encompass demographic and sociocultural mea­ federal definitions of rurality, including RUCC (USDA, 2013), RUCA
sures. However, we recognize no single measure of rurality, even a (USDA, 2010), and U.S. census designated urbanized areas (U.S. Census
multi-dimensional measure, can realistically account for all measures Bureau, 2019) and describe cases where CARR provides contrasting
that may be of interest to everyone (Woods and Heley, 2017). Moreover, results with these established definitions.
we view this proposed measure of rurality as a practical tool policy­
makers, researchers, and governmental agencies may utilize to study the 2. Data and methods
relationship between rurality and important socioeconomic outcomes
such as economic development or educational attainment and to target 2.1. Data sources and types
limited resources in areas to affect substantial changes (Agarwal et al.,
2009; Argent, 2008; Gallant and Robinson, 2011; Waldorf, 2006). In To develop a continuous operationalizable multi-dimensional mea­
short, though we recognize the complexity of rurality as a concept, we sure of rurality, we obtained spatial data on community services, ame­
focus on fundamental components of rurality from a sustainable rural nities, and relative rurality from national-scale publicly-accessible
development perspective, particularly those components critical to databases. To provide the greatest feasible level of spatial detail for
modern lives and that can be more easily influenced or acted upon services and amenities, exact georeferenced locations (unit-scale) were
(Woods and Heley, 2017). prioritized over aggregated data. Wherever available, official U.S. fed­
While demographic characteristics and sociocultural values are also eral datasets, such as those found at the Homeland Infrastructure
important to the rural space, they can be challenging to directly act upon Foundation-Level Data (HIFLD) Open Data portal or U.S. Census Bureau,
by interventions and governmental actions. For example, to address were obtained. These federal datasets were supplemented with national
disparities in food access across racial categories, a demographic char­ extracts from the community-driven, global open data project, Open
acteristic, a common intervention is to create a food bank (a service Street Map (OSM) (Haklay and Weber, 2008). OSM data extracts for the
associated with a piece of built infrastructure) (FeedingAmerica, 2022). entire U.S. were carried out in R using the package osmdata (Padgham
Moreover, in a post-industrial society, the economy and economic op­ et al., 2017). Relative rurality components, such as population esti­
portunities are no longer solely based on how far things are from urban mates, were obtained using aggregated census block-group data from
areas, and research has begun to examine how access to services such as the 2019 U.S. Census Bureau’s American Community Survey 5-year
broadband can bring jobs to rural communities (Galloway, 2007; estimates as provided by the National Historical Geographic Informa­
Townsend et al., 2013). As such, we view socioeconomic indicators as tion System (NHGIS) database (Manson et al., 2019).
outcomes that may be associated with rurality, but that are not a Measures of both availability and accessibility were constructed for
fundamental component of rurality. Hence, we focus on services and all services and amenities at block-group scale. We consider availability
amenities as core components of rurality in addition to issues of popu­ as the existence or area-based density of an asset and accessibility as the
lation density and remoteness. shortest distance to an asset. These two dimensions of asset provisioning
Services, such as supermarkets, general practitioner services, represent the presence of assets, which is necessary but not sufficient to
schools, banks and post offices, are critical to our daily lives and their ensure asset use, and proximity to the assets, which has been shown to
absence severely limits what we can access and what we are able to do. influence utilization of, for example, health, transportation, and green­
Similarly, while amenities may not be considered critical to daily lives space resources (Estabrooks et al., 2003; Gebremariam et al., 2017;
for some, they are central to what many consider modern life (e.g., Ozmen-Ertekin et al., 2007).
entertainment, restaurants). Limited access to many services and ame­ The measures of availability provide an indication of the supply of
nities is highly linked to rural communities and may negatively influ­ assets at the disposal of communities without considering their distri­
ence important outcomes such as educational attainment, healthcare bution across a block-group or the ease or convenience with which they

324
K.S. Nelson and T.D. Nguyen Journal of Rural Studies 97 (2023) 322–333

may be accessed. Simply considering availability can underrepresent distance to metro, and percent developed landcover) we have a total of
actual provisioning and use of resources in some block-groups without eight variables to construct the CARR index. Previously, a main con­
an asset within its boundaries, but where residents may have convenient struction method is to use either factor analysis or linear combinations
access to an asset that is located in a neighboring block-group. There­ of these components (Cloke, 2006; Mao et al., 2015; Nelson et al., 2021;
fore, an accessibility measure was calculated for each unit-level asset. Smith and Parvin, 1975; Zhao et al., 2019). We apply both of these
Accessibility was calculated as the shortest straight-line (Euclidean) approaches on the eight constructed variables. In terms of linear com­
distance from the centroid of a block-group to an asset. Larger values of binations, we could either use equal weighting of each of the three main
the accessibility measure indicate that an asset is located far from a components (availability, accessibility, or relative rurality) or we could
block-group and hence is less convenient to reach and make use of. give more weight to one set of components than another. Given that we
The measure of relative rurality based on distance was calculated as have no a priori reason to give more weight to one dimension over
the shortest straight-line (Euclidean) distance from the centroid of a another, we give equal weight to each of the three dimensions: avail­
block-group to the nearest major metropolitan area. The percent ability, accessibility, and relative rurality.
developed area measure was calculated using the 2016 National Land Using factor analysis on the eight standardized variables suggests a
Cover Database (Dewitz, 2019) where areas classified as low intensity, single factor with eigenvalue of 3.84. The factor loadings are all above
medium intensity, and high intensity in a block-group were considered 0.20 except for population size with a loading of 0.08. As such, we
to be developed. Population-based measures are readily available as construct a factor score using the seven standardized variables with
aggregated data at the block-group scale from the U.S. Census Bureau. loadings of more than 0.20 to create our first CARR index, dropping
All spatial analyses used to construct the CARR variables were con­ population size from the analysis. The Cronbach alpha of the final factor
ducted in R using the sf and raster packages (Hijmans et al., 2015; is 0.85, indicating a very good fit. The loadings on availability, popu­
Pebesma, 2018). lation density, and percent developed land were negative, and the
Following the calculation of each measure a complete dataset of each loadings on the other four variables were positive, resulting in a scale
availability and accessibility measure for all types of services and where higher scores indicate more remoteness and less access and
amenities and each relative rurality measure was compiled for all availability to assets. This CARR scale has a mean of zero, a standard
220,334 block-groups in the United States. For more information, see deviation of 0.696, a minimum-maximum of − 8.249 and 8.379, and is
the Supporting Information. approximately normally distributed. To make this scale more inter­
pretable, we rescale the minimum-maximum to 0 and 1, respectively,
2.1.1. Availability and accessibility factors where 0 represents very urban with extremely high availability and
We constructed 70 availability variables, 70 accessibility variables, accessibility to assets and 1 represents very remote and rural with
and 4 measures of relative rurality. With so many variables for avail­ extremely low availability and accessibility to assets. This rescaled fac­
ability (likewise, accessibility), it was necessary to reduce the number of tor is our main CARR index with a mean of 0.496, a standard deviation
variables contributing to this dimension and to eliminate non- of 0.042, and a range from 0 to 1 inclusively (Fig. S1).
meaningful variables. As such, we employ factor analysis on the avail­ Next, we create a second CARR index by using an “across-compo­
ability and accessibility measures independently and find one factor for nent” weighting scheme where we apply equal weights on each of the
availability and three factors for accessibility. (See factor analysis sec­ three components of availability, accessibility, and relative rurality. To
tion in Supporting Information and Table S3, which lists the specific make the two methods of factor analysis and linear combinations com­
variables and their loadings onto each factor.). We find a single factor for parable, we opt to exclude population size from the linear combination.
availability where larger values indicate more urban areas with a high The inclusion of the population variable does not substantively change
density of services and amenities (Fig. S4). the results. Within each component, we also give each variable an equal
In terms of accessibility, or how far away a service or amenity is weight; availability gets a weight of one, each of the three accessibility
located, we have three distinct factors. The first accessibility factor in­ factors gets a weight of one third, and each of the relative rurality var­
dicates that, for many communities, urban amenities such as arts cen­ iables gets a weight of one third. Specifically, we use the following
ters, movie theaters, hardware stores, libraries, and restaurants, tend to formula:
be located near each other or are similarly distant from block-groups.
1
Larger values of the first accessibility factor indicate lower access CARR = − Avail1 + (Access1 + Access2 + Access3 )
3
(greater distance) to non-essential amenities and recreation/entertain­
1( )
ment opportunities, while smaller values indicate high accessibility that − Popdensity − Distmetro + Landpercdeveloped
3
is often associated with the urban core (Fig. S5). The second factor in­
cludes more essential services such as childcare centers, places of We note that higher values of the availability factor, population
worship, banks, fire stations, and schools, many of which are more density, and percent land developed indicate more urban areas or hav­
uniformly distributed across space as they are essential for meeting basic ing more assets available, and as such, they are multiplied by negative
needs. Larger values of the second accessibility factor indicate lower one so higher values for the index indicate more remoteness or having
access (greater distance) to standard/essential community services. The less availability to assets. With this equal weighting scheme, the mean of
spatial distribution of this factor suggests an association with more this linear combination index is 0, with a standard deviation of 2.0,
suburban and residential areas (Fig. S6). The third factor consists of ranging from − 45.091 to 19.763. We also rescale this second CARR
places equidistant from both urban centers and natural resource index to have a range of 0–1 where 1 means most remote with the least
extraction areas, so some urban amenities and infrastructure and natural availability and accessibility to assets. The mean of the rescaled linear
resources are nearby. This third accessibility factor is spatially consistent combination index is 0.695 and the standard deviation is 0.032. This
with peri-urban areas, and larger values indicate lower, but approxi­ second CARR measure is mostly normally distributed with a left skew
mately equal, access to non-essential amenities and natural resource (Fig. S8).
extraction facilities (Fig. S7). In rough terms, we can think of these To compare the two versions, we first examine their correlation.
accessibility factors as places that have access to urban amenities, They are correlated at 0.96, indicating high correlation and suggest that
essential services, and peri-urban places. either method creates a similar measure. We prefer the factor analysis
CARR measure for three reasons. First, factor analysis provides a more
2.1.2. Constructing the CARR index robust statistical method that accounts for how the eight variables are
With our construction of availability and accessibility factors and correlated with each other and how they vary in their contribution to the
four measures of relative rurality (population size, population density, final index, while equal weights explicitly assume that each variable

325
K.S. Nelson and T.D. Nguyen Journal of Rural Studies 97 (2023) 322–333

within a component contributes the same amount.3 Second, the range of


the factor-score CARR index prior to rescaling is smaller and it is more
normally distributed than the equal-weight version. Third, when both
indices are rescaled to have a range of 0–1, the mean of the factor-score
CARR index is 0.496, which is nearly at the midpoint of 0 and 1, and is
only slightly skewed to the right, while the equal weight version’s mean
is 0.696 and is heavily skewed to the left. This difference in the means
and skewness would suggest that many more block-groups and counties
would be classified as urban and with high availability and accessibility
to assets with the equal-weights index relative to the factor-score index.
Furthermore, in practice, a mean value of about 0.5 provides a more
intuitive interpretation where values above the mean of the scale can be
generally seen as more “rural” and those below the mean as more
“urban”.

3. Results Fig. 1. Spatial distribution of CARR. Values of CARR for block-groups across
the coterminous U.S.
The developed CARR index is a continuous multi-dimensional mea­
sure of rurality constructed from 143 variables extracted from publicly near the middle of the range (near average for the entire country) tend to
available spatial datasets that accounts for availability and accessibility have a relatively low density of services and amenities. In Nevada, Nye
of services and amenities as well as traditional metrics of relative County (Fig. 2b) is made up of ~20 block-groups with a CARR range of
rurality such as remoteness and population density. Our preferred 0.53–0.99, where block-groups nearer to Las Vegas have relatively low
version of this index ranges from 0 to 1, where 0 represents “very urban” CARR scores and the highest CARR score in the entire US is observed in
with extremely high availability and accessibility to services and ame­ the mountainous region of the county to the north of Las Vegas. In
nities and 1 represents “very rural” with extremely low availability and contrast to the high variability counties in New York and Nevada where
accessibility to services and amenities. This distribution of the index is a relatively smooth spatial gradient from low to high values of CARR is
approximately normal, with a mean of approximately 0.5 (0.496) seen, Ventura County near Santa Barbara, CA (Fig. S2) has a minimum
(Fig. S1). We note that we are not advocating for a label of “urban” or value of 0.45 and a max of 0.85, where the only value above 0.5 is found
“rural” using the mean as the threshold as doing so would collapse the in a single block-group on an island off the coast.
distribution of the index into just a binary measure. As this is a contin­
uous measure, it is more accurate to think of higher values of the index 3.1.1. Differences and similarities with U.S. Census and USDA
as locales that have lower availability and accessibility to services and When comparing CARR to the U.S. rural-urban census designation,
amenities and for lower values of the index, more availability and we find, as expected, rural designated areas tend to be associated with
accessibility. However, having the mean of 0.50 provides a fortuitous higher CARR values (Fig. 3). While the average CARR value for rural
intuitive interpretation of index values, where values above the mean of designated places is ~0.55 the full range is from ~0.47 to ~1 displaying
the scale can be generally seen as more “rural” and those below the mean the great breadth of variation in provisioning of assets (services and
as more “urban.” amenities) across “rural” places. Interestingly, while urban designated
areas tend to have lower CARR scores the average is just ~0.49 and the
3.1. Spatial patterning complete range is from 0 to 0.8. While some of these “urban” designated
places are for block-groups in which only a small fraction of the block-
Generally, the pattern of CARR across the U.S. reflects patterns of group lies within a census “urbanized area,” there are still large num­
development, with large cities and densely populated areas having low ber of block-groups with a majority of area in an urbanized area with
values and sparsely populated areas with smaller towns having larger relatively high CARR scores. Block-groups in which more than 50% of
values (Fig. 1). The highest values of CARR are observed in the western the block-group area is within a census designated urban area have an
Great Plains region, interior of the Rocky Mountains, and in the arid average CARR value of 0.48 with a range from 0 to 0.77. These locations
desert of Nevada and southeast California. The lowest values of CARR may have a relatively high population and a large amount of developed
are observed along the northeast coast of the U.S. in Maryland, New land, yet few nearby services and amenities.
York, and Massachusetts. Counties in the U.S. with the highest CARR When compared to RUCC classification, we generally see, as the
values do tend to occur in “stereotypical” rural areas such as the arid RUCC classification tends towards more “rural” conditions, CARR value
mountain west, the wilderness of Montana, or small islands off the coasts generally increases (Fig. 4, See Table S2 for RUCC class descriptions.).
or in the Great Lakes. Relatedly, counties with low CARR scores tend to However, there is significant overlap in CARR across RUCC categories.
be located in or near major metropolitan areas such as New York City, In particular, RUCC class-6 areas have an average CARR value of 0.54
Boston, and San Francisco. that is lower (indicating more urban) than for the preceding class areas
The higher spatial resolution of CARR relative to other existing which have an average CARR value of 0.55. This is likely due to counties
measures of rurality highlights the heterogeneity present in provisioning with RUCC class-6 being located adjacent to a metro area, and hence
of services and amenities and relative rurality (Fig. 2). For example, New having greater access to services and amenities despite their lower
York County, NY (Fig. 2a) contains more than 100 populated block- population. Similarly, RUCC class-8 has an average CARR value of 0.56
groups with a minimum value of 0 and maximum value of 0.49. while class-7 has an average CARR value of 0.57. This reversal of some
Block-groups in this county with a low score tend to have very high of the CARR values associated with what are generally considered to be
population density and mostly developed land cover with a very high more rural categories reflects the greater emphasis of CARR on existing
density of services and amenities, while block-groups with CARR scores community physical assets over population size. In addition, we observe
there are more block-groups in areas designated as more urban using
RUCC and that there is more variability in CARR values in these areas
3
We also considered applying an “across-variable” weighting scheme where than in areas designated as more rural using RUCC.
equal weights of 1/7 are applied to each of the seven variables, and the results Similarly, we observe, as RUCA classification tends towards more
are substantively similar to the “across-component” weighting scheme. “rural” conditions, CARR values generally increase (Fig. 5). However,

326
K.S. Nelson and T.D. Nguyen Journal of Rural Studies 97 (2023) 322–333

Fig. 2. Counties with large CARR ranges. (a) New York County, NY and (b) Nye County, NV.

there is even more significant overlap in CARR across RUCA categories. CARR value indicating more “urban” conditions creating a low value
In particular, RUCA classes 4, 6, 7, and 9 have average CARR values outlier in our comparison of CARR and RUCA. While these cases
lower than for preceding classes which are expected to be more “urban”. certainly do exist, the greater bulk of outliers tend to exist on the
The description of these RUCA categories (see Table S2) suggests, rela­ opposite end of the CARR spectrum, where places categorized as “more
tive to the size and population density of the community, greater com­ urban” by RUCA have a higher (more “rural”) CARR value.
munity assets are available in communities with low levels of
commuting outside the community. Stated otherwise, communities with 3.1.2. Descriptive cases
greater access to services and amenities are associated with reduced We select and discuss several cases that contrast the differences be­
commuting rates, presumably because greater options for employment, tween the CARR index and current measures of rurality. More specif­
schooling, and other resources are available within their community. ically, we select block-groups categorized as metro or urban by either
For example, small towns with low commuting (class 9), on average, RUCC, RUCA or census that have the highest CARR index ratings, and
have nearly the same CARR as micropolitan areas with high commuting block-groups categorized as very rural by RUCC, RUCA or census that
(class 5). Trends in the number of block-groups classified as more urban have the lowest ratings on the CARR index.
or rural using RUCA follow the same pattern as observed for RUCC, and First, we contrast cases where the CARR index for the block-group is
we generally observe a greater range of CARR values in RUCA urban high, but it is categorized as urban or metro by RUCC, RUCA, or census.
areas than in RUCA rural areas. This comparison again points to a broad The first example is block-group 300859400015 in Montana which has a
range of levels of provisioning of community resources even within CARR score of 0.73 (recall that the mean for the US is ~0.5 with a
census tracts (which are, on average, ~27x times smaller than counties standard deviation of ~0.04) but is located in a census designated urban
but ~3 times larger than block-groups) classified as “urban”. area with nearly 100% of the block-group characterized as urban. This
Note that while we expect development occurring since 2010 (when CARR value reflects the large distance to a metropolitan area and rela­
the RUCA categories were developed) to be reflected in our CARR index tively high distances to amenities such as a community center, play­
which uses more contemporary data, this would be expected to lead to ground, or movie theater and to some services such as emergency
cases where areas classified as “more rural” according to RUCA have a medical service stations (Fig. 6). In this case while the CARR value

327
K.S. Nelson and T.D. Nguyen Journal of Rural Studies 97 (2023) 322–333

Fig. 3. Distribution of CARR for census designated urban and rural areas. Fig. 5. Distribution of CARR across RUCA classes. Violin plots showing the
Horizontal boxplots show the median value as a thick vertical line, the inter­ distribution of CARR across RUCA classes with the count of block-groups in
quartile range in the box, and the 95% confidence interval as whiskers. The each category (n). RUCA class 99 represents “unclassified” census tracts.
density of block-groups in each category (n) is displayed in the legend.
relatively urban conditions. However, the CARR for this block-group is
0.66 (more than three standard deviations above the mean) due to a low
density of community assets in the block-group and low access to
amenities such as libraries, museums, restaurants, and public parks
(Fig. S3).
In a different scenario, block-group 160739502003 on the edge of
Owyhee County, Idaho is counted as RUCC class-2 metro. However, this
is a large block-group in the “Four Corners” region, with an estimated
area of 2,850 square miles (much of which is designated wilderness
areas), an estimated population of only 602 in 2019, and little to no
access to important services and amenities. The distance from the center
of the block-group to the nearest grocery store is nearly 100 km (na­
tional average is 6 km with a standard deviation of 12 km) and the
distance to the nearest electric transmission line, bank, fire station, or
public school is far above national averages ranging from 55 to 100 km
(Fig. 7a). The CARR value for this block-group is 0.79 suggesting a high
degree of rurality instead of being classified as urban. Interestingly, this
block-group is classified as RUCA class-10 so CARR seems to agree well
with RUCA in this case, but not with RUCC. Similarly, block-group
320310035011 is categorized as a class-3 metro by both RUCC and
RUCA since it is a part of Washoe county in Nevada (where Reno, NV is
located). However, this block-group with a CARR of 0.74 is largely a
desert with large, designated wilderness areas and a 2019 population of
325 people in a 4,500 square mile block-group on the outskirts of the
Reno metro area. This block-group also has little access to essential
services such as supermarkets, fire stations, hospitals, and public schools
Fig. 4. Distribution of CARR across RUCC classes. Violin plots show the which are 50–90 km from the center of the block-group (Fig. 7b). In this
distribution of CARR scores along the y-axis, where areas of greatest width case CARR contrasts with both RUCC and RUCA.
indicate a larger percentage of records with similar CARR values and vertical
Next, we examine block-groups that are categorized as very rural and
lines indicate the full range of values observed. Count of block-groups in each
non-metro by RUCC and RUCA but do not have high CARR values. The
category (n) is provided along the x-axis.
first example is block-group 530559604002, a block-group in San Juan
County, Washington that is classified as a non-metro area that is
contrasts with the census designation, it is largely consistent with RUCC
completely rural with less than 2,500 people and not adjacent to a metro
and RUCA classifications for this block-group in the moderately rural
area. However, the population density for this block-group is not low
class 7 category (Table S2). In another example where CARR contrasts
since in 2019 it had 918 people in only 1.54 square miles. Moreover, it
with census, RUCC, and RUCA classifications, block-group
has access to many essential services and amenities such as banks, child
360450608041 in Fort Drum, New York is census-designated urban
care services, a museum, restaurants, supermarkets, and parks, within
with ~95% urban area with RUCC class-3 and RUCA class-1, indicating

328
K.S. Nelson and T.D. Nguyen Journal of Rural Studies 97 (2023) 322–333

Fig. 6. Map of block-group 300859400015 at Wolfs Point, Montana. The block-group boundaries are outlined in black and assets (supermarket, public school,
fire station, bank, library, playground, community center, museum, restaurant, movie theater, hospital) located within 20 km of the block-group are shown as points
with different symbols. (Assets not pictured are not available within 20 km)

Fig. 7. Map of block-group 160739502003 in Idaho (a) and 320310035011 in Nevada (b). Assets located within 50 km of the block-group are shown as points
with different symbols.

the block-group (Fig. 8). For example, the distance from the center of the by RUCC class-3. We argue this difference is untenable and confusing as
block-group to the nearest supermarket, hospital and public school is they provide non-comparable information. However, the CARR index
less than one km while amenities such as a movie theater, library, indicates that, while this block-group is part of a small town surrounded
museum, and park are less than two km away. While this block-group is by large forested areas and farms, it is about average in terms of rurality.
distant from a major metropolitan area, the moderate population den­ This bock-group has a relatively high population density of 1,826 people
sity, high percent land developed, and strong degree of access and per square kilometer in 2019 with moderate access and availability to
availability of many services and amenities lead to a CARR value of many community assets (Fig. 9a). The CARR value for this block-group is
0.505, which is slightly higher than the mean value of 0.496. This 0.472, which is slightly lower than the mean value nationally. In another
suggests that while this block-group is not a metro area with an abun­ case where CARR provides a more intuitive sense of rurality, we have
dance of services and amenities available to its citizens, it is also not a block-group 360470702021, which is on the edge of Kings County, New
“rural” area with little to no access to modern services and amenities. York. It is classified as rural by census but as a highly urban metropolitan
Next we examine block-groups where current categorizations among area by RUCC (class 1) and RUCA (class 2). It is classified as metro by
RUCC, RUCA, and census differ, but where the CARR index, arguably, RUCC and RUCA because it is a part of the New York City metro area, but
provides a better sense of rurality. For instance, block-group it is counted as rural by census because it is largely undeveloped land (a
181199558003 (and adjacent block-group 181199558004) is part of a marsh and airfield near Marine Park) with a very small population of 8
town in Owen County, Indiana, classified as rural by census but as metro people in 2019. We similarly argue that it is problematic that the current

329
K.S. Nelson and T.D. Nguyen Journal of Rural Studies 97 (2023) 322–333

Fig. 8. Map of block-group 530559604002 in Washington. Assets located within 100 m of the block-group are shown as points with different symbols.

Fig. 9. Map of block-group 181199558003 in Indiana (a) and 360470702021 in New York (b). Assets located within (a) 100 m of the block-group or 2 km (b)
are shown as points with different symbols.

categories would provide inherently very different information about 4. Limitations


this place, but also that none of the categories captures both the rural
(low population, low level of development, low density of services and There are three main limitations to the CARR index. First, similar to
amenities) and urban (near/in a metro area, high access to services and existing classifications of rurality, there are issues with the use of
amenities) characteristics of the block-group Fig. 9b). The CARR value administrative boundaries for delineation across space. In particular,
for this block-group is 0.485, reflecting both its location on the edge of block-group (also census tract and county) sizes are irregular, where
New York City and undeveloped character, but also its relatively strong areas with lower populations tend to be much larger than block-groups
access to many community assets nearby. with a high population density. This variability inherently alters esti­
In short, we illustrate how current categorizations can provide sub­ mation of distances and density of services and amenities, resulting in
stantially different, and sometimes conflicting, information about many averages across block-groups with populated and unpopulated areas as
areas in various contexts in the United States. On the other hand, we opposed to values that reflect the nature of specific populated places.
suggest that using the CARR index provides a more consistent and Despite this limitation, our block-group resolution index provides
intuitive characterization of rurality. greater clarity on differences across populated places than has

330
K.S. Nelson and T.D. Nguyen Journal of Rural Studies 97 (2023) 322–333

previously been made available for the entire conterminous U.S. Second, further allow for policymakers to address differential needs within a
despite our use of the most granular and up-to-date information, there single community, such as the issue of disparities in access to healthy
are still data limitations. In particular, infrastructure data are limited, food and healthcare (Rosik et al., 2021; Walker et al., 2010). Since CARR
especially longitudinally and in terms of quality attributes (e.g. road is measured at the block-group level, it can be examined in both “urban”
condition, hospital capacity) of specific assets. In our case, we assume and “rural” larger areas such as at the county or census tract level.
that all infrastructure reported in the datasets currently exist in a useable Fourth, since CARR is continuous and therefore more sensitive than
condition. Realistically, there is a likelihood that many assets (such as categorical measures to small changes, we will be more able to consis­
grocery stores) have closed doors and are no longer functioning. Simi­ tently examine how rurality is changing over time due to shifts in pop­
larly, there is a likelihood that many assets that currently exist have not ulation density or the availability of services and amenities. Fifth, CARR
yet been reported to the data providers. We also rely on OSM is a continuous measure of rurality that allows for greater nuance and
community-sourced data, so some infrastructure may be underrepre­ more granular analyses than most existing categorical measures. As a
sented where community members are less involved in digital mapping continuous measure, CARR can be used flexibly in modern regression
activities. However, our use of factor analysis on a great many asset analyses. For instance, CARR can be divided into deciles or interacted
types ensures that errors in one particular asset type will not strongly with other continuous and categorical variables to examine the inter­
bias the availability and access measures developed. Relatedly, even in action effects of rurality and other factors in regression analyses. Sixth,
cases where, for example, a store has closed doors, the presence of the the inclusion of community assets with traditional measures of rurality
physical store building represents a community asset that may be in CARR provides an intuitive and consistent measure of rurality as the
leveraged in future economic growth. Third, the CARR index does not descriptive case studies illustrate.
support sociological and cultural perspectives as no single measure of We see several potential applications of CARR and directions for
rurality can fully encompass everything. We also do not include de­ future research. First, to enable future work we will make the index
mographics outside of population density as part of our rurality measure freely and publicly available at a dedicated and persistent website.
since demographic profiles may shift rapidly due to migration trends or Second, temporal updates to the CARR index can be provided to examine
environmental and societal changes (e.g., climate change, pandemic how community assets and relative rurality may shift over time. Third,
responses) and are also less actionable and more difficult to address CARR may be constructed at different spatial resolutions (e.g., county
directly via policy actions. level, census tract, gridded). Fourth, scholars can create international
versions of CARR for many countries using OSM data (e.g., Nirandjan
5. Discussion and conclusion et al., 2022). Fifth, scholars can examine the relationship of CARR with a
host of important outcomes, including education, health, economics,
As previous work has recognized the limitations of heavily relying on politics and many other social outcomes. Sixth, policymakers and
population density and distance to metropolitan areas to measure governmental agencies may examine how community resources (and
rurality (Lichter and Brown, 2011; Minore et al., 2008; Nelson et al., investment in community resources such as highway and broadband
2021; Waldorf, 2006), our work addresses this fundamental issue by infrastructure) can lead to improved outcomes in areas such as educa­
suggesting a new measure of rurality that includes multiple dimensions tional attainment and health. They may also estimate the costs needed to
of rurality, including typical measures of remoteness and population elevate CARR in underserved areas to be on par with non-disadvantaged
density as well as the physical infrastructure that reflects advantages communities by investment in much-needed community assets, as well
and disadvantages in the provision of services and amenities critical to as estimate how regional investment strategies could reduce these in­
the maintenance of the population and economy of communities vestment needs.
(Agarwal et al., 2009; Binswanger and Deininger, 1997; Matson et al., In short, we make several contributions to the scholarly and policy-
2016; Siddiqi and Collins, 2017). To create a multi-dimensional measure relevant work around rurality with the CARR index. We envision the
of rurality, we build a unique dataset that links several national-scope construction and availability of the CARR index to have many implica­
public datasets at the most granular scale available to date. In our tions for policy and practice and open many new directions of future
analysis, we describe the spatial patterning of CARR across the U.S. and research in fields ranging from healthcare to geography to economics
the substantial variations that exist within counties. We also compare and to enable new developments in rural policy development and
differences and similarities of CARR against U.S. Census and USDA evaluation.
classifications. In particular, we select and discuss several examples that
contrast differences between the CARR index and current measures of Author contributions
rurality and illustrate how current classifications may provide substan­
tively conflicting information about the community while CARR is more Both authors contributed equally to this paper. Conceptualization:
consistent and intuitive. KSN, TDN. Methodology: TDN. Investigation: KSN, TDN. Visualization:
We suggest the CARR index has several strengths relative to previous KSN. Supervision: KSN, TDN. Writing—original draft: KSN, TDN. Wri­
measures of rurality. First, conceptually, CARR addresses the support of ting—review & editing: KSN, TDN.
basic life functions and well-being by considering the provisioning of
services and amenities for modern life (Gillespie-Marthaler et al., 2019). Data availability
The index includes many services and amenities that address the
fundamental needs that people have before they can focus on higher Data will be shared via the author’s github account at https://github.
level needs (Huitt, 2017). Second, since CARR focuses on com/katesnelson/CARR.
semi-permanent assets of communities, policymakers and governmental
agencies can identify communities with resources that can be leveraged Appendix A. Supplementary data
for growth and community sustainability (Matson et al., 2016; Siddiqi
and Collins, 2017). Third, researchers have argued that broad aggre­ Supplementary data to this article can be found online at https://doi.
gated spatial consideration of access to services would miss considerable org/10.1016/j.jrurstud.2022.12.025.
heterogeneity that exists within aggregated areas and that we need to
consider finer spatial scales of how services and amenities are distrib­
uted (Fortney et al., 2000a, Fortney et al., 2000b; Hall et al., 2006;
Minore et al., 2008). Along this line, the higher spatial resolution of
CARR acknowledges heterogeneity within communities, which would

331
K.S. Nelson and T.D. Nguyen Journal of Rural Studies 97 (2023) 322–333

References Kulig, J.C., Andrews, M.E., Stewart, N.L., Pitblado, R., MacLeod, M.L., Bentham, D.,
Smith, B., 2008. How do registered nurses define rurality? Aust. J. Rural Health 16
(1), 28–32.
TIGER/Line Shapefiles (Machine- Readable Data Files)/Prepared by the U.S. Census
Li, Y., Long, H., Liu, Y., 2015. Spatio-temporal pattern of China’s rural development: a
Bureau (TIGER), 2019. https://www.census.gov/geographies/mapping-files/time-
rurality index perspective. J. Rural Stud. 38, 12–26.
series/geo/tiger-line-file.2019.html.
Lichter, D.T., Brown, D.L., 2011. Rural America in an urban society: changing spatial and
Agarwal, S., Rahman, S., Errington, A., 2009. Measuring the determinants of relative
social boundaries. Annu. Rev. Sociol. 37, 565–592.
economic performance of rural areas. J. Rural Stud. 25 (3), 309–321.
Manson, S., Schroeder, J., Van Riper, D., Ruggles, S., 2019. IPUMS National Historical
Argent, N., 2008. Perceived density, social interaction and morale in New South Wales
Geographic Information System: Version 14.0 [Database]. IPUMS, Minneapolis, MN.
rural communities. J. Rural Stud. 24, 245–261.
Mao, L., Stacciarini, J.M.R., Smith, R., Wiens, B., 2015. An individual-based rurality
Ashley, C., Maxwell, S., 2001. Rethinking rural development. Dev. Pol. Rev. 19 (4),
measure and its health application: a case study of latino immigrants in North
395–425.
Florida, USA. Soc. Sci. Med. 147, 300–308.
Baker, E., Schmitz, D., Wasden, S., MacKenzie, L., Morris, B., 2011. Assessing Critical
Matson, P., Clark, W.C., Andersson, K., 2016. Pursuing Sustainability: a Guide to the
Access Hospital (CAH) Assets and Capabilities for Recruiting and Retaining
Science and Practice. Princeton University Press.
Physicians: the North Dakota CAH Community Apgar Program. Center for Rural
Meynard, J.M., Charrier, F., Le Bail, M., Magrini, M.B., Charlier, A., Messéan, A., 2018.
Health, The University of North Dakota, School of Medicine and Health Sciences.
Socio-technical lock-in hinders crop diversification in France. Agron. Sustain. Dev.
Retrieved from. https://ruralhealth.und.edu/projects/community-apgar-project/p
38 (5), 1–13.
ublications.
Minore, B., Hill, M.E., Pugliese, I., Gauld, T., 2008. Rurality Literature Review: Prepared
Berenguer, J., Corraliza, J.A., Martin, R., 2005. Rural-urban differences in environmental
for the North West Local Health Integration Network. Centre for Rural and Northern
concern, attitudes, and actions. Eur. J. Psychol. Assess. 21 (2), 128–138.
Health Research. Lakehead University.
Binswanger, H.P., Deininger, K., 1997. Explaining agricultural and agrarian policies in
Mountrakis, G., AvRuskin, G., Beard, K., 2005. Modeling rurality using spatial indicators.
developing countries. J. Econ. Lit. 35 (4), 1958–2005.
In: Proceedings of the 8th International Conference on GeoComputation.
Brown, D.L., Cromartie, J.B., 2017. The nature of rurality in postindustrial society. In:
Mueller, J.T., McConnell, K., Burow, P.B., Pofahl, K., Merdjanoff, A.A., Farrell, J., 2021.
New Forms of Urbanization. Routledge, pp. 269–283.
Impacts of the COVID-19 pandemic on rural America. Proc. Natl. Acad. Sci. USA 118
Brown, D.G., Johnson, K.M., Loveland, T.R., Theobald, D.M., 2005. Rural land-use trends
(1).
in the conterminous United States, 1950–2000. Ecol. Appl. 15 (6), 1851–1863.
National Rural Health Association Policy Brief, 2015. Quality of Life Impacts the
Cairns, R.C., 2014. Climate geoengineering: issues of path-dependence and socio-
Recruitment and Retention of Rural Health Care Providers. 1-6. Retrieved from http
technical lock-in. Wiley Interdisciplinary Reviews: Clim. Change 5 (5), 649–661.
s://www.ruralhealth.us/getattachment/Advocate/Policy-Documents/QualityofLi
Caschili, S., De Montis, A., Trogu, D., 2015. Accessibility and rurality indicators for
feRecruitmentRetentionProvidersFeb2015.pdf.aspx?lang=en-US.
regional development. Comput. Environ. Urban Syst. 49, 98–114.
National Rural Health Association Policy Brief, 2017. Community health workers:
Chen, H., Tang, L., Qiu, Q., Wu, T., Wang, Z., Xu, S., Xiao, L., 2018. Coupling between
Recommendations for Bridging Healthcare Gaps in Rural America. 1-22. Retrieved
rural development and ecosystem services, the case of Fujian Province, China.
from https://www.ruralhealth.us/NRHA/media/Emerge_NRHA/Advocacy/Policy%
Sustainability 10 (2), 524.
20documents/Community-Health-Workers_Feb-2017_NRHA-Policy-Paper.pdf.
Cloke, P., 2006. Conceptualizing rurality. In: Cloke, P., Marsden, T., Mooney, P.H. (Eds.),
Nelson, K.S., Nguyen, T.D., Brownstein, N.A., Garcia, D., Walker, H.C., Watson, J.T.,
Handbook of Rural Studies. Sage, London, pp. 18–28.
Xin, A., 2021. Definitions, measures, and uses of rurality: a systematic review of the
Cloke, P., Marsden, T., Mooney, P. (Eds.), 2006. Handbook of Rural Studies. Sage.
empirical and quantitative literature. J. Rural Stud. 82, 351–365.
Crouch, M., Nguyen, T.D., 2021. Examining teacher characteristics, school conditions,
Nguyen, T.D., 2020. Examining the teacher labor market in different rural contexts:
and attrition rates at the intersection of school choice and rural education. J. Sch.
variations by urbanicity and rural states. AERA Open 6 (4), 2332858420966336.
Choice 15 (2), 268–294.
Nirandjan, S., Koks, E.E., Ward, P.J., Aerts, J.C., 2022. A spatially-explicit harmonized
Doogan, N.J., Roberts, M.E., Wewers, M.E., Tanenbaum, E.R., Mumford, E.A.,
global dataset of critical infrastructure. Sci. Data 9 (1), 1–13.
Stillman, F.A., 2018. Validation of a new continuous geographic isolation scale: a
Ozmen-Ertekin, D., Ozbay, K., Holguin-Veras, J., 2007. Role of transportation
tool for rural health disparities research. Soc. Sci. Med. 215, 123–132.
accessibility in attracting new businesses to New Jersey. J. Urban Plann. Dev. 133
Ellis, F., Biggs, S., 2001. Evolving themes in rural development 1950s-2000s. Dev. Pol.
(2), 138–149.
Rev. 19 (4), 437–448.
Padgham, M., Lovelace, R., Salmon, M., Rudis, B., 2017. osmdata. Journal of Open
Estabrooks, P.A., Lee, R.E., Gyurcsik, N.C., 2003. Resources for physical activity
Source Software 2 (14).
participation: does availability and accessibility differ by neighborhood
Pebesma, E.J., 2018. Simple features for R: standardized support for spatial vector data.
socioeconomic status? Ann. Behav. Med. 25 (2), 100–104.
Rev. Javer. 10 (1), 439.
FeedingAmerica, 2022. Find Your Local Food Bank. https://www.feedingamerica.
Ramos, F.R., Uitermark, J., 2021. An introduction to DUIA: the database on urban
org/find-your-local-foodbank.
inequality and amenities. PLoS One 16 (6), e0253824.
Fortney, J., Rost, K., Warren, J., 2000a. Comparing alternative methods of measuring
Roscigno, V.J., Crowle, M.L., 2001. Rurality, institutional disadvantage, and
geographic access to health services. Health Serv. Outcome Res. Methodol. 1 (2),
achievement/attainment. Rural Sociol. 66 (2), 268–292.
173–184.
Rosik, P., Stępniak, M., Wiśniewski, R., 2021. Delineation of health care deserts using
Fortney, J., Rost, K., Warren, J., 2000b. Comparing alternative methods of measuring
accessibility measures: the case of Poland. Eur. Plann. Stud. 29 (6), 1151–1173.
geographic access to health services. Health Serv. Outcome Res. Methodol. 1 (2),
Scottish Executive, 2006. Scottish Index of Multiple Deprivation 2006 Technical Report.
173–184.
Edinburgh. Office of the Chief Statistician, Scottish Executive. https://www.gov.scot
Gallent, N., Robinson, S., 2011. Local perspectives on rural housing affordability and
/publications/scottish-index-multiple-deprivation-2006-technical-report/.
implications for the localism agenda in England. J. Rural Stud. 27 (3), 297–307.
Sher, J.P., 2019. Education in rural America: A reassessment of conventional wisdom.
Galloway, L., 2007. Can broadband access rescue the rural economy? J. Small Bus.
Routledge. https://doi.org/10.4324/9780429047411.
Enterprise Dev. 14 (4), 641–653.
Siddiqi, A., Collins, R.D., 2017. Sociotechnical systems and sustainability: current and
Galluzzo, N., 2018. A preliminary quantitative analysis of rural development in Romania
future perspectives for inclusive development. Curr. Opin. Environ. Sustain. 24,
using the PLS-SEM. Albanian J. Agric. Sci. 17 (3), 125–133.
7–13.
Gebremariam, M.K., Vaqué-Crusellas, C., Andersen, L.F., Stok, F.M., Stelmach-
Smith, B.J., Parvin, D.W., 1975. Estimating the relative rurality of US counties. J. Agric.
Mardas, M., Brug, J., Lien, N., 2017. Measurement of availability and accessibility of
Appl. Econ. 7 (2), 51–60.
food among youth: a systematic review of methodological studies. Int. J. Behav.
Townsend, L., Sathiaseelan, A., Fairhurst, G., Wallace, C., 2013. Enhanced broadband
Nutr. Phys. Activ. 14 (1), 1–19.
access as a solution to the social and economic problems of the rural digital divide.
Gillespie-Marthaler, L., Nelson, K., Baroud, H., Abkowitz, M., 2019. Selecting indicators
Local Econ. 28 (6), 580–595.
for assessing community sustainable resilience. Risk Anal. 39 (11), 2479–2498.
Uhl, J.H., Hunter, L.M., Leyk, S., Connor, D.S., Nieves, J.J., Hester, C., et al., 2022. Place-
Gold, A., Burnstein, E., Scally, C.P., Su, Y., 2021. Reenvisioning Rural America: How to
level Urban-Rural Indices for the United States from 1930 to 2018. arXiv preprint
Invest in the Strengths and Potential of Rural Communities.
arXiv:2202.13767.
Haklay, M., Weber, P., 2008. Openstreetmap: user-generated street maps. IEEE Pervasive
United Nations, 1999. World Urbanization Prospects: 1999 Revision. United Nations,
computing 7 (4), 12–18.
New York.
Hall, S.A., Kaufman, J.S., Ricketts, T.C., 2006. Defining urban and rural areas in US
Waldorf, B.S., 2006. A continuous multi-dimensional measure of rurality: Moving beyond
epidemiologic studies. J. Urban Health 83 (2), 162–175.
threshold measures (No. 379-2016-21891). https://ageconsearch.umn.edu/reco
Henderson, J., 2002. Building the rural economy with high-growth entrepreneurs. Econ.
rd/21383/.
Rev. Fed. Reserv. Bank Kans. City 87 (3), 45–75.
Waldorf, B., Kim, A., 2015. April. Defining and measuring rurality in the US: From
Henning-Smith, C., 2020. The unique impact of COVID-19 on older adults in rural areas.
typologies to continuous indices. In: commissioned paper presented at the Workshop on
J. Aging Soc. Pol. 32 (4–5), 396–402.
Rationalizing Rural Area Classifications. Washington, DC. https://sites.nationalacade
Higgs, G., 1999. Investigating trends in rural health outcomes: a research agenda.
mies.org/cs/groups/dbassesite/documents/webpage/dbasse_167036.pdf.
Geoforum 30 (3), 203–221.
Walker, R.E., Keane, C.R., Burke, J.G., 2010. Disparities and access to healthy food in the
Hijmans, R.J., Van Etten, J., Cheng, J., Mattiuzzi, M., Sumner, M., Greenberg, J.A., et al.,
United States: a review of food deserts literature. Health Place 16 (5), 876–884.
2015. Package ‘raster’. R Package, p. 734.
Woods, M., Heley, J., 2017. Conceptualisation of Rural-Urban Relations and Synergies.
Huitt, W., 2017. Hierarchy of needs. In: Moghaddam, F. (Ed.), The SAGE encyclopedia of
University of Aberystwyth. ROBUST deliverable 1.1).
political behavior. SAGE Publications, Thousand Oaks, CA, pp. 356–357. https://doi.
Wunderlich, G.S., Rapporteur), 2016. Rationalizing Rural Area Classifications for the
org/10.4135/9781483391144.n166.
Economic Research Service: A Workshop Summary. The National Academies Press,
Inagami, S., Gao, S., Karimi, H., Shendge, M.M., Probst, J.C., Stone, R.A., 2016. Adapting
Washington, DC.
the Index of Relative Rurality (IRR) to estimate rurality at the ZIP code level: a rural
classification system in health services research. J. Rural Health 32 (2), 219–227.

332
K.S. Nelson and T.D. Nguyen Journal of Rural Studies 97 (2023) 322–333

Zhao, J., Ameratunga, S., Lee, A., Browne, M., Exeter, D.J., 2019. Developing a new USDA ERS, 2013. United States Department of Agriculture Economic Research Service.
index of rurality for exploring variations in health outcomes in Auckland and Rural-Urban Continuum Codes Documentation. https://www.ers.usda.gov/data-pr
northland. Soc. Indicat. Res. 144, 1–26. oducts/rural-urban-continuum-codes/documentation/. (Accessed 13 May 2022).
U.S.Dept. Of Health & Human Services. 2022. Guidance Portal - Defining Rural USDA ERS, 2010. United States Department of Agriculture Economic Research Service.
Population Accessed https://www.hhs.gov/guidance/document/defining-rura Rural-Urban Commuting Area Codes Documentation. https://www.ers.usda.gov/
l-population#:-:text=Office%20of%20Management%20and%20Budget%20Definit data-products/rural-urban-commuting-area-codes/documentation/. (Accessed 13
ion&text=All%20counties%20that%20are%20not,as%20either%20Metro%20or% May 2022).
20Micro. (Accessed 12 May 2022). Dewitz, J., 2019, National Land Cover Database (NLCD) 2016 Products: U.S. Geological
Survey data release, https://doi.org/10.5066/P96HHBIE.

333

You might also like