3
Other Rural Area Classification Systems Used in the United States and Internationally
This chapter summarizes the third workshop session, which introduced rural area classification as done elsewhere in the United States and internationally. The session began with a presentation by Brigitte Waldorf (Purdue University) of the commissioned paper Defining and Measuring Rurality in the United States: From Typology to Continuous Indices (Waldorf and Kim, 2015). Leif Jensen (Pennsylvania State University) described labor market area (LMA) delineations in the United States. Paolo Veneri (OECD) described the OECD rural classification system and the adaptation of that system in the European Union, and provided an example of a classification used in Italy. Keith Halfacree (University of Swansea) provided a social constructionist critique of rural area classification systems. Mark Partridge (Ohio State University) was the moderator.
STATEMENT BY BRIGITTE WALDORF1
Waldorf first introduced Ayoung Kim, who co-authored her presentation and the commissioned paper. Waldorf discussed in general terms what rurality is, and how it might be measured, coded, ranked, or delineated. She said that most definitions in the literature are either vague or nonexistent. As a result, analysts sometimes code and classify something that is not defined. However, analysts agree that rurality is a multidimensional concept, and that multidimensionality raises methodologi-
_______________________
1This presentation is based on Waldorf and Kim (2015).
cal issues and requires priority-setting and subjective decision making. Multidimensionality also raises questions about what these dimensions are, how they are measured, and what their relative importance is. For example, the Rural-Urban Continuum Codes (RUCC) have at least two dimensions, size and adjacency, calling for a subjective decision regarding their relative importance. In this case, priority was given to size and adjacency followed (see Chapter 2).
Methods to Measure, Code, and Rank Rurality
Waldorf stated that, in general, the existing methods for classifying multidimensional concepts can be divided into two groups: (1) typologies or classifications, which are further divided into either threshold-based or similarity-based; and (2) aggregate indices.
Threshold-Based Typologies
In the United States, almost all rural typologies use the Census Bureau’s delineation of urban areas, the core-based statistical areas of the Office of Management and Budget (OMB), or both. The rural typology suggested by Isserman (2005) is an example of a typology that utilizes the Census Bureau’s definition of urban areas but does not use the OMB delineation. An example of a typology that relies both on the Census Bureau and the OMB delineations is the Rural-Urban Continuum Code of the Economic Research Service (ERS). The rural-urban classification of the National Center for Health Statistics uses the OMB but not the Census Bureau delineations. Finally, an example of a typology that uses neither the Census Bureau nor the OMB delineation is the OECD typology for the United States, discussed in a subsequent presentation.
She noted that the Census Bureau and the OMB typologies are also used as criteria in rural-urban typologies for spatial objects different from counties. These typologies vary in terms of what they are classifying, such as blocks and tracts. The typology can also refer to a single location that can be classified as being rural or urban. Looking at the various criteria used in rural-urban typologies, size and density are the most frequently used dimensions.
Similarity-Based Typologies
As an alternative to using thresholds, similarity indices can be used to create a classification, Waldorf explained. The n objects—counties, tracts, or zip codes, for example—are assigned to k types, using m variables or criteria. Unlike with the threshold-based typologies, similarity-based
typologies group objects on the basis of similarity within this m-dimensional space. The m dimensions are defined by criteria such as size and density, and similarity is measured by distance in m-dimensional space. There are different ways of measuring distance, and the selection of the distance measure is a subjective decision. But once that subjective decision is made, then the classification procedure becomes very mechanical and data-driven. This method is rarely used, although it is easily implemented and can be applied to spatial objects of any scale.
Aggregate Indices
An alternative to typologies is the aggregate index. As Waldorf described, aggregate indices are similar to typologies in that they select k criteria or variables to capture multidimensionality. They are different because they do not require thresholds that divide the multidimensional space into discrete compartments but, instead, collapse the multidimensional space into one-dimensional space. Within the one-dimensional space, a rurality aggregate index would rank objects (counties or tracts, for example) by increasing rurality. As such, the aggregate index does not answer the question of what is rural versus urban, but is a relative measure that allows comparison of areas by their degree of rurality. It is responsive to changes in any of the underlying dimensions. She pointed out that both the threshold-based typologies and aggregate indices are based on very critical subjective decisions: the choice of thresholds in the case of threshold-based typologies, and the functional specification for the collapse of the multidimensional into one-dimensional space in the case of aggregate indices.
The United Nations designed a successful aggregate index with its Human Development Index (HDI). A similar procedure can be applied to create an aggregate index of rurality. Waldorf (2006, 2007a, 2007b) designed such a continuous threshold-free index of rurality, the Index of Relative Rurality, which ranges between 0 (most urban) to 1 (most rural) for 3,108 counties, excluding Hawaii and Alaska. She and Kim updated the Index of Relative Rurality for 2000 and 2010 for 3,141 counties, including Alaska and Hawaii. She said the design consisted of four steps: (1) identifying the dimensions of rurality—population size, density, built-up area, and remoteness; (2) selecting measurable variables to adequately represent each dimension—population size, population density, percent urban, and network distance to the closest metropolitan area; (3) rescaling the variables onto a comparable scale—the bounded interval [0,1]; and (4) selecting a function that links the rescaled variables so that multidimensionality is reduced into one-dimensionality—in the absence of theoretical guidance, the unweighted average was used.
Concluding Comments
Waldorf closed by noting most delineations of rural-urban areas use threshold-based typologies. In her view, thresholds create artificial similarities and dissimilarities, and, in most cases, threshold choices are not well justified. They are dependent on a fixed spatial scale to be classified. Every time a different spatial scale is used, different thresholds are needed. However, she said, one advantage of threshold-based classifications is their simplicity.
An alternative is the Index of Relative Rurality, an aggregate index such as the HDI of the United Nations. Compared to threshold-based typologies, its main advantages include being threshold-free, continuous, and scale-independent. Moreover, it is a relative measure so that spatial objects can be ranked by their degree of rurality, and even subtle changes in the underlying dimensions over space and time can be revealed. Finally, the index is analytically more easily handled than categories of a typology. With all these advantages, the Index of Relative Rurality is a useful addition to the set of existing threshold-based classifications, but it is not a substitute, Waldorf said.
STATEMENT BY LEIF JENSEN
Jensen introduced his collaborators Danielle Rhubart and Chris Fowler. He acknowledged a new cooperative agreement with ERS, as well as National Institutes of Health support for the Population Research Institute at Pennsylvania State University and his own involvement with Regional Research Project W-3001, which is looking at the impacts of the Great Recession on small town and rural area demographic change.
Labor Market Areas—Definitions and Methodology
Jensen focused on LMAs, defined as economically integrated geographic areas within which individuals can reside and find employment within a reasonable distance, or can readily change employment without changing their place of residence. An alternative definition is a geographic area encompassing both the place of work and the place of residence of a local population.
LMAs were first considered in the early to mid-1980s with a collaborative project between the Rural Labor Market Section of ERS and Regional Research Project S-184, “Labor Markets and Labor Differentiation in Nonmetropolitan America.” He noted that project had considerable leadership from Charles Tolbert, then at Florida State University, and Calvin Beale at ERS (Tolbert and Killian, 1987).
Key outcomes from the project were the delineation of 382 LMAs and the preparation of the PUMS-D by the Census Bureau. PUMS-D was a public-use microdata sample from the 1980 decennial census long-form questionnaires that contained individual-level and household-level data. For each individual it identified the LMA of residence. In a follow-on Regional Research Project S-229, “The Changing Structure of Local Labor Markets in Nonmetropolitan Areas,” many of the same researchers conducted research on LMAs and arranged for an updating with the 1990 decennial census data that was called the PUMS-L.
The rationale for this effort was recognition of the inadequacy of individual counties as units to understand an area’s economy, Jensen said. As stated by Tolbert and Killian (1987, p. 1), “A local economy and its labor force are not bounded by the nearest county line, but by interrelationships between buyers and sellers of labor”: hence, a labor market. These researchers also recognized the inadequacy of nonmetro areas in capturing the diversity of rural America. Jensen said they were motivated by the need for a geographic standard to capture labor markets, and, more generally, to better understand the implications of context for individual outcomes, and in particular the effects of characteristics of labor markets on the circumstances of people living within them.
The methodology that they followed drew on counties and county equivalents as the building blocks, using journey-to-work data and population size of these places. Subsequently, ERS subdivided LMAs into metro and nonmetro. The method relies on proportional flow measures—basically, the total number of commuters exchanged by two counties divided by the size of the smaller county’s labor force. He said this approach emphasizes the reciprocity between counties rather than assuming that nonmetro counties and outlying areas rely solely on metro areas. These proportional flow measures go into a symmetric matrix of proportional flows, basically, a county-to-county flow matrix, which is then subject to hierarchical cluster analysis. He said this is an iterative process of aggregating counties based on the strength of their relationship of the proportional flows of commuters.
The process aggregates counties into one large cluster. A dendrogram is used to decide where to stop the process and, thus, where the clusters best define labor markets. This process also led to the formation of commuting zones. Jensen pointed out that one of the goals of the Regional Research Projects was to work with the Census Bureau to produce special sub-samples. For example, the PUMS-D had LMAs within them, but because the Census Bureau was sensitive to making sure that these areas had at least 100,000 residents in order to protect the confidentiality of the census data, a last step in the process was to aggregate commuting zones into somewhat larger LMAs. He noted Beale was involved in this part of the process.
Research Applications
Jensen said that regional research projects and other users of the PUMS-D and the subsequent PUMS-L were interested in the effect of context, notably, the implications of rural LMA characteristics for individual outcomes. They examined relationships between LMA characteristics and such variables as returns to human capital, gender and household labor supply, determinants of off-farm employment, income packaging among the poor, and race-ethnic differences in unemployment. These kinds of questions remain important in view of changing rural landscapes, Jensen pointed out.
Jensen stated a goal of their current project is to review, replicate, and evaluate the prior methods; to update the LMA delineations with more recent data; and ultimately to design a new set of functional LMAs that reflect current population settlement and commuting using the most appropriate recent data. They are replicating the past work but with more advanced hardware and software, and will be able to analyze the entire flow matrix for all counties simultaneously and take advantage of other software advances. They are updating with contemporary data from the American Community Survey (ACS), but also exploring the use of the Longitudinal Employer-Household Dynamics (LEHD) Origin-Destination Employment Statistics (LODES2) data.
Jensen noted one drawback of the current methodology is that it was based entirely on the decennial census, conducted in April every 10 years.3 This raises issues of seasonality that perhaps the ACS can overcome. On the other hand, the ACS has a smaller sample size and greater margins of error. Because of the 10-year cycle of the census, the data cannot capture economic cycles, a limitation that using ACS data may also address. Finally, the current approach results in monocentricity or places being dominated by large counties. Jensen and his colleagues plan to solve this problem by replacing the proportional flow method and continuing with hierarchical agglomerative clustering methods. He observed that there are a variety of ways to describe the connection between two counties. He noted that proportional flows is a good measure, but it emphasizes connections to counties with large workforces and builds a monocentric representation of commuting patterns.
_______________________
2Available: http://lehd.ces.census.gov/data/ [October 2015].
3Journey to work data are now only available from the ACS.
Future Research and Implications for Defining Rurality
Jensen highlighted the questions he and his colleagues hope to illuminate in their research project:
- How does the geography of LMAs change using constant methodology applied across years?
- How have the characteristics of urban and rural LMAs changed over time, using the most recent LMA definition applied to previous years?
- How have the implications of LMA context for individual-level outcomes changed over time?
Jensen said that using data available through Pennsylvania State University’s Census Research Data Center (RDC) will provide access to internal Census Bureau versions of data files to see how commuting has changed within LMAs independent of how places are grouped. They will evaluate within-LMA commuting patterns of different population groups by type of employment.
While this research has been to define LMAs, Jensen said it has implications for defining rurality. He observed the persisting need to understand rural labor markets outside of metropolitan areas, and that using LMAs can help to appreciate this rural diversity. He said he hopes an analysis of the richer data available in the Census Bureau’s RDC will allow for a better understanding of the links between rural and urban labor markets and will also support exploration of some of the limitations of the ACS. More generally, Jensen indicated that he and his colleagues are interested in whether “rural” can be defined more accurately and more meaningfully using subcounty delineations.
STATEMENT BY PAOLO VENERI
Veneri described how rural area classification is done outside the United States, focusing on Europe and other countries in the OECD. He noted that reflecting on the definition of rural is very timely because the OECD is also currently discussing how to update its own urban-rural classification, how to improve the classification with new data, and how to address emerging issues. Veneri described the OECD regional typology, including extensions to the typology by the European Union, and provided an example of a different classification approach used by Italy.
The OECD Official Databases
The OECD publishes databases at three spatial scales that are validated by the 34 OECD countries and national statistical offices. Most of the data are at the national level, called Territorial Level 1 (TL1). A regional database relates to large regions and small regions. Large regions (TL2) are usually the first government layer after the national/federal one, such as U.S. states. The small regions (TL3) often correspond to administrative entities. In the United States, small regions are Economic Areas as defined by the Bureau of Economic Analysis.
The current OECD classification of rural areas is applied to TL3 regions, but the building blocks used to build the classification are “local administrative units,” such as counties, wards, or municipalities. A population density criterion identifies three categories of regions: predominantly urban, predominantly rural, and intermediate. The predominantly rural regions are further divided into those that are close to a city or rural remote, based on distance to cities, Veneri explained.
OECD Method
Veneri reiterated that local administrative units are the building blocks of the OECD method. Local units are counties in the United States, wards in the United Kingdom, and usually municipalities in other countries. The issue of comparability is key because local units in different countries may have very different sizes. However, this compromise was reached in order to have an international urban-rural classification adopting a consistent method.
The OECD method starts by classifying local units along the urban-rural continuum, as follows:
- Local units are classified as rural if their population density is below 300 inhabitants per square kilometer.
- Regions are classified as rural based on the proportion of the population living in rural local units. If the proportion is higher than 50 percent, then the region is classified as predominantly rural.
- Regions are classified as intermediate if the proportion of the population living in rural local units is between 15 and 50 percent.
- Finally, regions are classified as close to a city if driving time to a city is less than one hour for at least 50 percent of the population.
Adjustments are made based on the size of city: a region becomes intermediate or urban if it contains a city of at least 200,000 or 500,000 inhabitants, respectively.
Revising the OECD Classification and the European Union Approach
Veneri said that the OECD is currently discussing how to update and revise its territorial classification for two main reasons. First, there are purely statistical needs to update the current classification, based on 2001 census data, using the most recent population census data for all countries. Second, a revision could make use of currently available information, methods, and data, such as high-resolution grid cells and Geographic Information System elaborations, that might support a more precise and comparable urban-rural classification across OECD countries.
He noted that the first consideration in revising the classification system is to ask why the OECD classifies space along the urban-rural continuum. The main reason is to compare territories across countries for research purposes, which requires comparability across countries. A second objective is to have policy-relevant units of analysis within each country, to the extent possible.
As one option, he said the OECD could adopt the extension proposed by the European Union, already in use. In most European countries, the new rural codes coincide with OECD TL3 regions. The main innovation used in constructing the EU classification is the use of grid cells of one square kilometer as building blocks, instead of local administrative units. Another important difference is how the proximity to a city is measured. The EU approach uses the OECD definition of Functional Urban Areas,4 which are also consistent across countries, to take into account the presence of cities and the distinction between the rural regions close to a city and remote rural regions.
The EU methodology classifies clusters of contiguous 1-km2 cells along the urban-rural continuum according to their population density, Veneri said. The classified cells can then be used to classify other geographies of interest, such as local units (municipalities, counties, etc.) or regions. In other words, he said, the use of grid cells allows comparability across countries to be maximized (grid cells have the same 1-km2 across all countries) and the result can be applied for the classification of any larger region of interest.
Cells are classified based on their population density and size into one of three types: urban centers (or high-density clusters), urban clusters, and rural grid cells. The first two are basically the urban space. The urban centers are those clusters of contiguous cells with high density of at least 1,500 inhabitants per square kilometer, and 50,000 in population size. Urban clusters are groups of contiguous cells with a population density higher than 300 inhabitants per square kilometer. Finally, rural grid cells
_______________________
4See OECD (2012).
are cells not included in the other two types. This method has not been applied to the United States, only to Europe.
Veneri said the main advantage of this method is that the classification is fully comparable across all countries that have data about the population in each grid cell. Additionally, the method can be used to create classifications for any larger geography of interest, whether relevant for policy purposes, or for research and analysis.
The Example of Italy
Veneri described a different classification system used in Italy as an example of the classification of rural areas based on policy objectives.
In Italy, “Inner Areas” are territories characterized by “a not adequate access to essential services to assure a certain level of citizenship among population.”5 The intention is to measure lack of access to health care, education, and transportation. This classification is driven by policy purposes: It supports a policy package to foster local development and improve opportunities by improving access to services.
“Service Centers” are defined as municipalities that have inside their territories an exhaustive range of secondary schools, at least one highly specialized hospital, and a railway station, approximating the presence of minimum services for education, health care, and transport. All Italian municipalities have been classified according to the distance (travel time) from these Service Centers. For example, if this distance is higher than 75 minutes, they are identified as Inner Areas, which are the object of these policies.
Concluding Comments
Veneri said the extension of the OECD urban-rural classification proposed by the European Union could be fairly easily applied in the United States. It would allow consistent comparisons with other developed countries. He observed that population density probably remains the most straightforward criterion on which to base urban-rural classifications, especially when these are done across different countries or for very large and different territories. However, he observed the Italian Inner Areas example demonstrates that other classifications are possible based on proximity criteria only. He said this can make sense when the objective of the classification is for policy purposes rather than solely for analysis.
_______________________
5See Ministry of Economic Development (Italy) (2014).
STATEMENT BY KEITH HALFACREE
Halfacree presented a critique of the idea of producing rural classifications. He noted producing these classifications is a useful exercise as long as the context in which it is undertaken is recognized.
Halfacree explained rural geographer Paul Cloke constructed an index of rurality for England and Wales based on census data (Cloke, 1977). It produced a plausible and understandable map of rurality in England and Wales. London and the southern cities of Wales stood out as very urban, and the most remote areas in Wales, southwest of England, and the north of England stood out as rural. The classification using census data was very much in the spirit of the times, Halfacree said, and was plausible and influential. However, Cloke repudiated his index (Cloke, 1994), stating that it was an inappropriate way of addressing the idea of what and where is rural and that selecting a number of variables to represent “the rural” predetermined the outcome. According to Halfacree, Cloke and others said they made an assumption that “the rural is there” and all that was needed was to find the right sort of measures to express it, instead of considering exactly what was meant by rural.
Recognizing “Rural” as Socially Constructed
Halfacree recognized the central importance of classification, categorization, or taxonomy as a central practice of human life. However, he said, it deserves critical reflection.
People produce categories, he said. Taxonomic practices or putting things into boxes minimizes ambiguity and vagueness in the world, brings things into the open, and provides simple, clear, communicable, controlled consensus on meaning (Bourdieu, 1990). But, he said, taxonomies are not without problematic aspects.
Another term used to articulate what is involved within taxonomic practices is the process of discrimination. Here, Halfacree said, “natural” may be distinguished from “more-than-natural” entities. Discriminating natural phenomenon such as species of animals or plants may be fairly easy. However, doing so for more-than-natural entities, such as rural taxonomy, is more challenging.
Halfacree considered the rural as a social representation of space, noting that Moscovici (1984) talked about dealing with the world’s perpetual complexity by simplifying and formalizing into social representations. He said social representations have three functions: to organize, understand, and interact with the world. All are part of the central cognitive stages of everyday life.
In earlier work (Halfacree, 1993), he said he argued that rural could be
seen as one example of social representations, namely a social representation of space. He cited Copp (1972) for describing rural as an important rhetorical device of intractable popular significance. In terms of a social representation, Halfacree said the rural according to Moscovici’s theory would have a relatively stable core, a figurative nucleus, and more transient associated elements. It would be a combination of concrete images and abstract concepts. He said that Moscovici saw these social representations as social, very much encultured, and not inherently as individual constructs.
To apply a social construction and social representation perspective, he identified two principal tasks. The first is to discover how space is socially represented among a particular group of people or in a region to find out how they imagine the space around them. The second task is to discover place and prominence of the rural within this process, using ordinary language that indicate which categories stand out, how people imagine and talk about the space around them, and how they organize it and use it. He asked a series of questions: Where does the “rural” fit in that? Who shares each of these social categories? If the rural does feature strongly, does it feature for everyone or does it feature for particular groups in society? Which social delineation (class, ethnicity, etc.) seems to be most important?
It is also important to think about historical and geographical variations, Halfacree stated. Historical variation is often captured by culture and cultural change. How does rurality, or people’s ideas about rurality, change over time? How does the concept of rurality vary between different countries, for different types of physical geography, or different physical climactic environments?
Questions About Rural in the United States
Halfacree questioned how the concept of rurality in the United States is based on the impact of British ideas of rurality. Second, he asked about the validity of seeking one way to define rural. Is there a reason to make allowances for regional variations, he asked, noting his question also applies to the OECD quest for global definitions of rural. His third critical question related to the legitimacy or value of the term “rural” beyond the academic or policy arena. In other words, how much do people on the street recognize and use the term? There may be a need to investigate this assumption through surveys and other methods, he suggested. His fourth critical question was, if rural is accepted as a key concept, what is its figurative nucleus? He questioned the place and importance today of more traditional ideas of the rural such as agriculture, isolation, and
remoteness versus more “novel” elements such as use in leisure and recreation, amenity value, and scenic beauty.
In closing, Halfacree stated that he would argue that the social construction and the social representation idea of the rural has proved very influential, certainly in British rural studies and beyond from the 1990s into the 2000s. As a result, he said he believes the concept of “the rural” as something socially constructed by groups and from experiences is important. However, the current trend is to consider the rural a bit more broadly. The physical reality of the world also needs to be taken into consideration when constructing any indices or measures of rurality. However, he concluded, that is beyond the scope of the present talk.
OPEN DISCUSSION
Mark Partridge (Ohio State University) opened the discussion with a suggestion directed to Jensen. He said if LMAs are defined and ERS puts them online, people will use them; they will not look at the documentation or caveats. Recently, he said, people who do not normally conduct regional and urban research are using LMAs without understanding them. The other issue is that a classification system is designed to cover the whole territory, and each part of it has some number attached to represent its LMA. As a result, some marginal areas are included that have very little connection with the other parts of the region, especially at the edges. He suggested a special designation for areas with only a weak attachment to their assigned LMA as a way to remind users of the caveats.
John Pender (ERS) asked about dimensions that go into alternative classification systems, noting that Cromartie earlier summarized those dimensions for the ERS classification systems (see Chapter 2). There could be an aspect of validation involved with getting people’s opinions of “rural” to identify the kinds of dimensions that might be involved. He asked about other work validating the dimensions that might go into a classification scheme.
He noted that an issue related to thresholds is that it is not clear how they are set. He suggested that thresholds could be set to maximize explanatory power for a set of outcome variables, such as poverty or industrial composition, by comparing how well different threshold levels stack up in terms of the share of the variance that is explained by different thresholds. He suggested that a research opportunity might be to look at validation of thresholds so they do not seem arbitrary.
Waldorf observed that the most important underlying issue is the way dimensions are selected. They also seem to be selected in an arbitrary way, or at least one not perfectly justified, she said. Theoretically, there are an infinite number of typologies, with a typology for every additional
dimension and threshold. The question is what is rurality and how a definition can be validated without typologies. She said she believes it is impossible, but robustness checks can be made by following up with different types of rurality measures.
Michael Ratcliffe (Census Bureau) said Census Bureau staff are trying to understand the origin of the 2,500 threshold. At the end of the 19th century or early 20th century, Census Bureau geographers felt that places of 2,500 contained the kinds of services and functions typically urban at that time. He observed that he discussed with Cromartie, Fitzsimmons, and others whether 2,500 is still meaningful, noting 50,000 was used in some earlier definitions of urban areas in the early 20th century and was adopted as the starting point for cities and urbanized areas. In 1950, part of the decision about the threshold was driven by resources and limitations on the ability to manipulate data for the country. With automation, it would be possible to consider any threshold, but these thresholds are enshrined in programs and legislation. Every time they try to change a threshold, someone is impacted negatively, Ratcliffe said.
Michael Woods (Aberystwyth University) referred to Figure 5 of the workshop-commissioned paper (Waldorf and Kim, 2015, p. 16) in which the authors showed areas with decreasing rurality and areas with increasing rurality between 2000 and 2010. He said that his meta-analyses suggest that rural is becoming urbanized. He asked Waldorf if her areas of increasing rurality are another way of saying these areas are losing population. He asked whether changes over time indicate that areas are becoming more or less rural, or whether the nature of rurality is changing, which he said also links to Halfacree’s presentation.
David Brown (Cornell University) proposed two reasons for partitioning geography and the population into urban and rural areas. One is for scholarly research, to figure out if, net of other factors, living in a place called “rural” makes a difference to outcomes such as poverty. The second reason is for policy, to determine if structural differences between areas require different service delivery approaches or a need for different eligibility criteria. He observed that a different way to phrase the question is, “Does rural make a difference either in certain outcomes or in terms of public policy?”
Brown referred to one of Halfacree’s questions about how a hybrid combination can underpin the necessary task of mapping rural. He asked if by hybrid, Halfacree meant a combination of structural approaches and a more social constructive approach. He went on to ask how these threads could be brought together. Halfacree responded that hybridity was partly as Brown described. The whole idea of validation and checking on the terms used is central. It is imperative to think carefully about defining “rural” areas. In that respect, many so-called definitions of rural
are constructed, he said. They are not really definitions of rural, but rather definitions of a particular problem, such as isolation or remoteness. A more comprehensive idea of rural space needs to make allowances for what people understand and utilize as the concept of rural space.
Robert Gibbs (ERS) commented that not only is the concept of rural socially constructed, but also the social construction is infused with power. If the social constructions of rural are infused with power, then those who develop classifications are exercising power, he said. The question is, what are the consequences, who gains and who loses? There are different ways the construction and the labeling of things as rural might have consequences, he said. He noted the purposes of classification had been discussed, but not the consequences, except in terms of providing health care facilities and such.
Tom Johnson (University of Missouri) commented that the discussion of dimensionality of rural suggests two research questions: what are the dimensions and what weight should be attached to each dimension? Researchers would not want to limit themselves either in dimensionality or in the weighting of those dimensions, he said. Instead, it is important to look at the relationship between density, remoteness, and the outcome variables. The concept of rurality is more one of communication and explication of these issues following the research.
Gregory Hooks (McMaster University) commented on the discussion of power and social construction. He said the first OECD classification would never apply to a country based on Anglo-Saxon settler societies, like Canada, the United States, or Australia. Rural does not mean the same in all places across a continent, and remote areas may be overrepresented with indigenous peoples, he stated. The idyllic image of what rural means may vary depending upon historical processes and their legacies. The variation in U.S. and world history, Hooks said, brings into sharp relief that very precise measures of rurality are not the same.
This page intentionally left blank.
