Toward Understanding Homelessness: The 2007 National Symposium on Homelessness Research. Rural Homelessness

By:
Marjorie Robertson, PhD, Alcohol Research Group, Albany, CA
Natalie Harris, Miami Valley Housing Opportunities, Dayton, OH
Nancy Fritz, MDiv, Governor's Office, Augusta, ME
Rebecca Noftsinger, Westat, Rockville, MD
Pamela Fischer, PhD, Substance Abuse and Mental Health Services Administration, Center for Mental Health Services, Rockville, MD

 

Disclaimer: This paper was developed for the National Symposium on Homelessness Research held on March 1-2, 2007. The Symposium was conducted by Abt Associates Inc. and Policy Research Associates Inc. under contract for the Office of the Assistant Secretary for Planning and Evaluation, U.S. Department of Health and Human Services; and the Office of Policy, Development, and Research, U.S. Department of Housing and Urban Development. The paper presents the views and opinions of the respective author(s) and does not necessarily represent the views, positions, and policies of the federal government.

"

Abstract

For the 1998 Symposium on Homelessness Research, rural homelessness was not assigned as a paper topic in its own right. Due to its increasing significance, the authors prepared a paper on rural homelessness for the 2007 Symposium. Given the somewhat limited formal research available, the authors supplemented their literature review with information from government documents and technical assistance materials as well as input from an expert panel of researchers and practitioners. The paper summarizes what is documented to date about the characteristics of people who are homeless in rural areas and examines whether rural homelessness and the service approaches to address it can be differentiated from urban homelessness. The authors identify gaps in current knowledge about rural homelessness and recommend new directions for research.

Introduction

Since the late 1970s when large numbers of homeless people began appearing on city streets, homelessness has been regarded by the public as primarily an urban phenomenon. Although advocates and providers in rural areas were also seeing increasing numbers of people without housing, prompting speculation about the existence of the hidden homeless in rural communities (Post, 2002), very few researchers addressed issues related to rural homelessness. This dearth of scholarly literature was deemed of sufficient concern to the Institute of Medicine (IOM) Committee on Health Care for Homeless People in preparing its comprehensive report on Homelessness, Health, and Human Needs (1988) that a special study of rural homelessness was commissioned (Patton, 1988). In 1991, the Housing Assistance Council (HAC) published Rural Homelessness: A Review of the Literature, summarizing the various studies documenting rural homelessness (Housing Assistance Council, 1991). By 1995, rural homelessness was a visible issue warranting focus from the U.S. Department of Agricultures Rural Economic and Community Development (RECD) and the Federal Interagency Council on Homelessness (ICH), which held a series of regional conferences with providers and advocates (Burt, 1996). A decade following the IOM report, the federal government convened the 1998 National Symposium on Homelessness Research to present papers on the state of the art of homelessness research, published as Practical Lessons (Fosburg & Dennis, 1999). However, although Rosenheck, Bassuk, and Salomon (1999) identified rural homelessness as contributing to the diversity of American homelessness, they did not discuss the topic as a phenomenon needing specific enquiry or presenting unique characteristics and challenges for service providers. Since the 1998 Symposium, most of the additional research conducted to address homelessness policy and practice has continued to focus on urban settings, with the result that rural homelessness remains much less well understood.

There are three main reasons for examining what is known about rural homelessness. The first is to document the prevalence of the problem. The second is to examine whether rural homelessness can be differentiated from urban homelessness as well as to identify differences in how persons who are homeless are served in rural communities. The third is to identify the gaps in current knowledge and recommend areas critical to new research. This paper addresses these issues, synthesizing research conducted over the past 25 years. Because little research has been done on rural homelessness, the literature reviewed was expanded beyond peer-reviewed journals to include government publications, technical assistance documents, and program materials encompassing rural homelessness and related topics (i.e., sources covering homelessness in general, rural poverty, rural health, rural mental health and substance abuse, as well as various data sources). In a further effort to fill gaps in the literature, a number of telephone interviews were conducted with service providers to inform the services component of the paper. In addition, this paper draws on an expert panel comprising researchers, service providers, government officials, and advocates convened in October 2006 to discuss cutting-edge issues in rural homelessness and to identify areas for future enquiry.

Prevalence of Rural Homelessness

The Impact of Definitions on Rural Homelessness Research

The lack of consensus on how to define rurality has impeded research into rural homelessness by making it difficult to specify the population studied in a consistent manner (National Institute on Drug Abuse, 1997). To date, federal agencies and researchers have not settled on a single definition of rural but rather construct definitions specific to various uses. Moreover, the definitions have changed over the years to reflect demographic shifts as well as changing notions of urbanicity. While definitions adopted by various government agencies tend to overlap, there are important distinctions in the geographic areas delineated as rural, affecting population estimates, services eligibility, and the like.

Rurality is typically defined in contrast to urbanicity. The most commonly used definitions, such as those developed by the U.S. Bureau of the Census, the Office of Management and Budget (OMB), and the U.S. Department of Agriculture (USDA), are based on population density and proximity to metropolitan areas. For example, prior to 2003, OMB defined a metropolitan community as a population nucleus with a population of 50,000 or more and the economically tied surrounding area. Communities with more than 5,000 but fewer than 50,000 people were designated as urban clusters. Subsequently, OMB added micropolitan communities with a population of up to 10,000 plus surrounding county areas where at least 25 percent of the population commutes to the micropolitan center. All other (or rural) communities are considered non-core. The U.S. Bureau of the Census definition of urban includes urbanized areas consisting of one or more central places and adjacent territory with a population density of at least 1,000 per square mile that together have a minimum residential population of at least 50,000 people, and urban clusters of densely settled areas having at least 2,500 but fewer than 50,000 people. Rural areas constitute all territory, population and housing units not classified as urban (Coburn, 2007). Because urban and rural classifications crosscut other geographic hierarchies, rural pockets can be located in both metropolitan and non-metropolitan areas.

Within the continuum of communities defined as rural, frontier areas represent the extreme. Although not a Census classification, frontier areas are usually defined as having a very low population density, typically fewer than seven persons per square mile (Popper, 1986). Most frontier counties are found in western states and are characterized by peoples relative isolation and dispersion across large geographic areas. Frontier areas account for about 400 communities countrywide, which cover about 45 percent of the countrys land mass. It is important to define frontier areas because of their distinctive characteristics relative to other rural areas; for example, economic downturns start earlier and snowball faster in frontier communities because they are less complex and often rely on a single industry (e.g., tourism, ranching, farming, logging, or mineral extraction) (Ciarlo et al., 1996).

Although these various agency definitions have been used, for example, to document differences in drug use and health across categories of population and geographic location, it is important to note that these definitions were developed to provide nationally consistent standards for collecting federal statistics for geographic areas and not for inappropriate uses such as determining program eligibility of individual applicants (Standards for defining Metropolitan and Micropolitan Statistical Areas, 2000). Nevertheless, these designations are frequently used to determine eligibility and distribute many types of federal funding  from homeland security to housing  in ways that exclude persons who are homeless in rural and frontier areas (Substance Abuse and Mental Health Services Administration, Center for Mental Health Services, 2006).

Equally important is how homelessness itself is defined in rural areas and how the unique circumstances of rural living affect the enumeration of people who are homeless in rural settings. The most widely used definition of homelessness for determining policy comes from the McKinney-Vento Homeless Assistance Act (1987), which defines a homeless person as, (1) an individual who lacks a fixed, regular, and adequate nighttime residence; and (2) an individual who has a primary nighttime residence that is  (A) a supervised publicly or privately operated shelter designed to provide temporary living accommodations (including welfare hotels, congregate shelters, and transitional housing for the mentally ill); (B) an institution that provides a temporary residence for individuals intended to be institutionalized; or (C) a public or private place not designed for, or ordinarily used as, a regular sleeping accommodation for human beings. Structures commonly found in rural settings that the U.S. Department of Housing and Urban Development (HUD) calls substandard but stable housing do not meet criteria stipulated in Part C of the definition (Substance Abuse and Mental Health Services Administration, Center for Mental Health Services, 2006). Although persons living in housing that has been condemned can be defined as homeless by HUD, a formal and consistent condemnation process does not exist in most rural communities. This means that a structure considered not fit for human habitation in Washington, D.C., would not be designated as such in Viper, Kentucky (Substance Abuse and Mental Health Services Administration, Center for Mental Health Services, 2006). Consequently, undercounts of rural homeless people may result from exclusion of persons living in substandard structures  structures that in an urban setting would be condemned. To counteract this problem, some rural communities have worked with their local county governments to create a special designation for these properties in accord with the residence criteria stipulated in Part C of the homelessness definition. A few are now using the BOCA building code (Building Officials and Code Administrators International, 1996) definition of not fit for human habitation to distinguish persons who are homeless from those living in substandard housing.

The definition of a chronic homeless person as "any unaccompanied homeless individual with a disabling condition who has either been continuously homeless for a year or more, or has had at least four episodes of homelessness in the past three years, created by HUD, the U.S. Department of Health and Human Services (HHS), the U.S. Veterans Administration (VA), and the U.S. Interagency Council on Homelessness (ICH) to determine eligibility under the consolidated plan (Consolidated Plan Revisions and Updates for Department of Housing and Urban Development, 2006), has been widely criticized for its exclusion of homeless families who have similar patterns of chronic homelessness (Child Welfare League of America et al., 2005). If families are a larger proportion of rural homeless populations than urban, use of this definition may also lead to disproportionate undercounts of chronic homelessness in rural relative to urban areas.

Enumeration of Rural Homeless People

Problems defining, locating, and sampling have made enumerating the homeless population virtually impossible (Cowan et al., 1988). In urban areas, estimates have commonly relied on counts of persons using services. However, by this measure, homeless persons in rural areas are likely substantially undercounted due to the lack of rural service sites, the difficulty capturing persons who do not use homeless services, the limited number of researchers working in rural communities, and the minimal incentive for rural providers to collect data on their clients (Burt et al., 1999). As a consequence, the number of homeless persons counted in a given area tends to correlate with how vigilant the surveyors are in finding them (Aron & Fitchen, 1996; Hudson, 1998). For example, when a man well-known by local providers to be chronically homeless was asked why he had not been included in a local HUD point-in-time count, he replied that he wasnt a dog that, when called, would come out to be counted (Substance Abuse and Mental Health Services Administration, Center for Mental Health Services, 2006).

Most estimates of homelessness in rural areas have been extrapolated from rates reported in urban areas. For example, in 1987, as part of a study of eating patterns of homeless people in U.S. cities with populations of 100,000 or more, the Urban Institute collected data that researchers used to create one-day and one-week estimates of the numbers of homeless persons in central cities in the United States. Although no small city, suburban, or rural locations were included in the initial survey, these estimates were then extrapolated to the rest of the United States using assumptions about the ratio of homelessness rates in central cities compared to other areas of the country. This method produced a one-week estimate of 229,000 persons from shelters and meal programs in cities of over 100,000 and an extrapolation to the whole country of 500,000 to 600,000 people (Burt and Cohen, 1989).

The next national study of homelessness was mounted in 1996. The National Survey of Homeless Assistance Providers and Clients (NSHAPC) was a collaborative effort of 12 federal agencies, coordinated by the Interagency Council on Homelessness. Urban Institute researchers conducted an analysis of interviews with people sampled from shelters, meal programs, and other homeless service agencies in communities as well as from some generic housing, community action, and welfare agencies in communities with few or no dedicated homeless services. Data from this study were used to estimate that 444,000 to 842,000 persons were homeless in the United States on any given night (Burt et al., 2001).

After weighting their data to be nationally representative of homeless assistance programs during an average week and to prevent double counting, NSHAPC estimated that about 9 percent of its homeless sample was drawn from rural areas in the U.S. (Burt et al., 1999). However, since people were only counted if they were clients in a broad array of targeted homeless service programs, the sample and findings of the study may have underrepresented the rural homeless population, because not all people who are homeless in rural areas use homeless service programs, and dedicated homeless assistance programs tend not to exist in smaller rural or frontier communities.

Another national one-day estimate of homelessness derives from the efforts of local communities to obtain counts of sheltered and unsheltered homeless people as part of their applications for funding under HUDs Supportive Housing Program and other related homeless assistance funding streams. Beginning in 2005, when HUD began requiring methodologically defensible counts or no counts, HUD continuum-of-care applications included systematic counts, implemented with varied levels of rigorous methodology, from most communities in the country. The National Alliance to End Homelessness (NAEH) summed the reported counts from 2005 to create an updated base estimate of persons without housing in the U.S. From the 2005 counts from more than 450 communities, the National Alliance estimated that 744,313 Americans experienced homelessness on any given night in 2005 (National Alliance to End Homelessness, 2007). Applying the NSHAPC rate of 9 percent to this most recent national estimate by the National Alliance suggests that at least 67,000 adults may have been homeless in rural areas of the U.S. on any given night in 2005.

Unfortunately, using urban rates to estimate the extent of rural homelessness may not provide an accurate count due to differences in population density and other factors. Although rural areas represent 75 percent of the countrys land mass, rural residents make up only 17 percent of the total population of the country (Johnson, 2006). While the absolute number of homeless persons in rural communities is smaller than that found in cities, the prevalence of homelessness (i.e., the number of homeless persons relative to the general population) has been estimated to be greater in some rural areas than in some major metropolitan areas (Kentucky Housing Corporation, 1994, 2002; Lawrence, 1995; Post, 2002). Furthermore, Burt (1999) cautioned that most studies that yield estimates of population size have methodological differences that make valid comparisons virtually impossible. One example of a survey that allows comparison, because it includes both rural and urban communities, is the statewide 1993 Kentucky Homeless survey, which found higher rates of homelessness in some rural communities compared to urban areas. The very rural Fleming and Lee counties had rates of 109 and 152 per 10,000 respectively while the most urban counties of Fayette and Jefferson had homeless rates of 32 and 17 per 10,000 (Kentucky Housing Corporation, 1994).

The numbers of homeless youth are not usually included in estimates of homeless populations in the United States. However, as part of the Youth Risk Behavior Survey sponsored by the Centers for Disease Control in 1992 and 1993, Ringwalt and colleagues analyzed data for a national representative household sample of 6,496 youth ages 12 through 17 (Ringwalt et al., 1998). Using a broad definition of homelessness, the researchers reported that overall, 7.6 percent of the national sample had experienced homelessness in the previous year. Contrary to findings reported for adults, the prevalence of homelessness among youth did not vary significantly by race, family poverty, family structure, or region of the country. The annual rate of homelessness for youth in the previous 12 months was 8.4 percent in the rural (or non-metropolitan) areas compared to 8.3 percent in the Metropolitan Statistical Areas (MSAs) with central cities and 6.8 percent in MSAs without central cities. This finding suggests that homeless youth are a sizable and important subpopulation among homeless persons in both rural and urban areas. In contrast to homeless adults, whose homelessness is more often related to structural or economic factors, homeless youth in rural areas may become homeless due to conflict with parents or other household members, eviction, and other personal circumstances, as do homeless youth in urban areas (National Coalition for the Homeless, 2006a; Robertson & Toro, 1999; Wilder Research Center, 1998).

Rural Poverty in the United States

Demographic changes in rural communities since the 1970s have contributed to persistent poverty, institutions and infrastructures being stretched to their limits, and escalating housing costs (Johnson, 2006), all of which contribute to homelessness in rural areas. From the 1920s through the 1960s, rural counties grew slowly through natural increase (i.e., more births than deaths), even though millions of people moved from rural areas to cities. Subsequently, rural areas have experienced growth as a result of a reverse in earlier migration trends, with people, including immigrants, moving from cities into rural counties (except in the heartland of the Midwest), with about 17 percent (50 million) of the total U.S. population currently residing in rural areas (Johnson, 2006).

Homelessness has been characterized as the extreme end of poverty (Hopper & Hamburg, 1986). Poverty rates in the United States are highest in remote rural counties and central cities (Center for Family and Community Life, 2005; Mosley & Miller, 2004; National Coalition for the Homeless, 2006b). In 2005, 15.1 percent of rural populations were living in poverty compared to 12.5 percent of non-rural populations (Jensen, 2006). Among the 500 poorest counties, non-metropolitan (i.e., rural) counties outnumbered metropolitan counties by 11 to 1, and 48 of the 50 poorest counties were in rural areas (Aron, 2004). Recent research indicates that the odds of being poor are between 1.2 and 2.3 times higher for non-metropolitan residents than for metropolitan residents (National Coalition for the Homeless, 2006b). While the majority of rural low-income people are non-Hispanic whites, ethnic minorities in rural areas are particularly disadvantaged relative to both rural non-Hispanic whites and urban non-whites (Jensen, 2006; Mosley & Miller, 2004). For example, in non-metropolitan areas in 2001, almost one-third of African Americans and one-quarter of Hispanics lived in poverty compared to 11 percent of non-Hispanic whites (Jensen, 2006). American Indians in rural areas are also disadvantaged. In Montana, 38.4 percent of American Indians were living at or below poverty levels compared to 12.7 percent of all white persons in the state (Montana Council on Homelessness, 2007). OHare and Johnson (2004) point out that the highest poverty rates for children are also found in rural areas. Meanwhile, anti-poverty programs are often implemented less successfully in rural communities than in urban communities due to factors such as lack of transportation, physical and social isolation, stigma attached to seeking government assistance, and a dearth of health care providers and facilities (OHare & Johnson, 2004).

While rural communities have higher homeownership rates than most urban communities, 24 percent of rural households are renters. Because the rural housing stock is generally of lower quality, rural renters are twice as likely to live in substandard housing as their urban counterparts (12 percent of rural versus 6 percent of urban renters) (Housing Assistance Council, 2003b). Moreover, the cost burden is higher for rural renters than urban renters due to lower incomes (averaging $20,500 in rural versus $36,800 in urban areas in 2003). Lower incomes result in 36 percent of rural renter households paying more than 30 percent of their adjusted income toward housing (Housing Assistance Council, 2003b). Meanwhile, federal funding for rural rental housing programs, including the U.S. Department of Agriculture (USDA) Rural Development Section 515 Program, has been drastically reduced, making it even more difficult to address this disparity. Under the Section 515 program, direct loans are made for terms of up to 50 years to for-profit developers, nonprofit corporations, and public bodies to construct, purchase, or rehabilitate rental housing in rural areas for low- and moderate-income families, elderly persons, and persons with disabilities; loans and grants also fund the development of housing for domestic farm laborers (Housing Assistance Council, 2007; National Rural Housing Coalition, 2004).

Additionally, unrealistically low fair market rents in rural communities do not create an incentive for housing development in rural areas. While there is a lack of affordable rental housing throughout the country (measured by the percentage of people paying over 30 percent of their income for housing) (Saulny, 2006), fair market rents continue to remain low in most rural communities. Fair market rents (FMRs) serve as the payment standard used to calculate gross rent estimates (i.e., rent plus utilities) under the Rental Voucher program for 354 metropolitan areas and 2,350 non-metropolitan county FMR areas (HUD Office of Policy Development and Research, 1995). As HUD is quick to point out, setting these standards means balancing between creating rent payments high enough to stimulate housing availability but low enough to serve as many persons as possible (U.S. Department of Housing & Urban Development, Office of Policy Development & Research, 1995).

In calculating FMRs, HUD works closely with the U.S. Bureau of the Census and takes into account both decennial census rent data and the Bureaus American Housing Surveys. This information is then supplemented with random telephone surveys. This process proves to be successful for most metropolitan communities in determining a base rent from what is being paid in the community. However, the process is less successful in rural communities because of the small populations and small stock of rental units in these communities. The American Housing Surveys are conducted only in the 44 largest U.S. metropolitan communities, and most rural communities do not have enough rental units available to make a reliable random telephone survey feasible. In order to address the issue of unreliable FMRs in rural communities, HUD has implemented minimum FMRs using the statewide average FMR of non-metropolitan counties. However, since these non-metropolitan county estimates are all considered to be low, the minimum FMRs actually affect very few communities and do not create high enough minimum rents to spur rental housing development in the countrys poorest rural communities.

Meanwhile, minimum operational costs required to manage rental housing do not vary by community size or wealth. While rural developers are remarkably creative in developing housing, the average one-bedroom operating cost in rental housing development in communities less than 20,000 was between $3,749 and $4,064, similar to the average urban annual operating expense of $3,800 figured by the Federal Home Loan Bank. Therefore, the minimum break-even monthly FMR for most low-income rental housing in rural areas was about $333 per unit, not much different from the corresponding break-even monthly FMR for urban areas. Yet, the monthly one-bedroom FMR for high growth, urban communities, including Washington D.C. at $1,134, New York City at $1,069, Los Angeles at $1,016, and San Francisco at $1,239, was much higher than the FMR in rural areas. The minimum FMR for the poorest rural communities was as low as $335 per month (U.S. Department of Housing & Urban Development, Office of Policy Development & Research, 1995).

Causes of Rural Homelessness

Homelessness among adults and families is not evenly distributed across rural areas (Aron, 2004; Post, 2002), but rather is concentrated in communities that have histories of persistent poverty; are primarily agricultural or have economies based on declining mining, forestry, or fishing industries; have reduced employment opportunities due to changing economies (e.g., due to the replacement of family farms with mechanized or corporate farms); or are economic growth areas that attract both more job seekers than can be absorbed and relatively high-income residents whose presence increases housing costs and other living expenses. In his report for the Institute of Medicine on rural homelessness, Patton (1988) wrote that like urban homelessness in the United States, rural homelessness is fundamentally due to the interaction of structural and personal factors. However, some of the structural factors for current rural homelessness have different historic roots than those for urban homelessness, stemming from the rural economic restructuring of the early 1980s that included twin recessions, massive farm foreclosures, and loss of labor-intensive rural manufacturing to foreign competition. The impact of the rural economic crisis was uneven and had the most devastating impact on counties that lacked economic diversification (i.e., in which local economy was dominated by a single industry) and on counties where rural poverty has historically been entrenched, such as the southeastern states. Patton (1988) wrote that of 231 counties that ranked in the bottom fifth for income over the previous 30 years, all but 18 were in the southeastern United States. He emphasized the concept of relative burden, by which he meant that relatively low numbers of homeless persons can easily overwhelm a rural communitys resources.

Today, structural factors that continue to contribute to rural homelessness include inadequate housing quality, declines in homeownership, rising rent burdens, and insecure tenancy resulting from changes in local real estate markets. Related factors include lack of infrastructure to support employment, such as child care and public transportation, and long distances between low-cost housing and employment opportunities. Inadequate treatment opportunities for disabling medical and behavioral health problems, including serious mental health and substance use problems, also contribute to vulnerability to homelessness in rural areas (Center for Family and Community Life, 2005; National Coalition for the Homeless, 2006b). On occasion, natural disasters (such as Hurricane Katrina) contribute to homelessness in rural areas through displacement of formerly housed persons. In addition, local and statewide studies report domestic violence as a major contributor to homelessness of women (Intergovernmental Human Services Bureau, 2003; Kentucky Housing Corporation, 1994; Montana Council on Homelessness, 2007), and unaccompanied youth (Wilder Research Center, 1998).

Characteristics of Rural Homelessness

Status of Research on Rural Homelessness

Relatively little research has focused specifically on rural homelessness, with extant studies based mainly upon descriptive surveys of clients (demographic and social characteristics) or service providers. Interviews of service providers or clients typically used convenience samples, and anecdotal comments or informal observations were often reported as though they had the weight of more controlled empirical findings. Much of the research is dated, with the first statewide studies of rural homelessness conducted in the mid-1980s. The Ohio Mental Health Study (Roth et al., 1985; Roth & Bean, 1986) was an ambitious project that initiated survey data collection in 1983. Researchers conducted a representative survey with 790 urban and 189 non-urban homeless people statewide: the urban sample comprised 81 percent of the respondents while the non-urban sample included respondents from both the mixed/urban (10 percent) and rural (9 percent) counties. Homelessness was defined broadly to include literally homeless adults (i.e., staying on the streets or in shelters) as well as residents of cheap hotels and motels and persons who were doubled up. In 1985, the Vermont Department of Social Welfare conducted a survey of district directors of state social welfare offices and other key informants to determine the number of homeless people in their areas, the dynamics of the problem, and service issues (Housing Assistance Council, 1991; Vermont Department of Social Welfare, 1985). A total of 2,800 persons were estimated to need shelter during 1984. In 1987, the California State Department of Mental Health conducted an enumeration and representative survey of homeless adults in three California counties, including one rural (Yolo) and two urban counties (Alameda and Orange ) (Vernez et al., 1988). Using a more restrictive definition of homelessness, homeless adults were recruited from dedicated service sites (e.g., shelters or meal programs) and from the streets in an attempt to collect a more representative sample than from service sites alone.

Recent studies have expanded the scope and improved methods of studying homelessness in rural areas, going beyond mere population descriptions toward identifying ways to serve the rural homeless population and to intervene with at-risk populations in rural areas (Aron, 2004). For example, in 2006, a sample of 3,582 adults was interviewed in a one-night statewide survey of homeless adults and unaccompanied youth in Minnesota using a broad definition of homelessness. Selection criteria for the study included adults and youth in shelters (emergency shelters, domestic violence shelters, and transitional housing); non-sheltered adults and youth sampled from meal programs, drop-in centers, bridges, encampments, and other sites in more than 80 cities, towns, and surrounding areas; and other non-sheltered adults and youth who had stayed one night or longer in a shelter or been literally homeless (i.e., on the streets, in a car, in an abandoned building, or some other place not meant for habitation) any time within the previous seven days. This definition was broader than most by sampling from street sites and domestic violence shelters and by including anyone who had been literally homeless in the previous seven days. About one-third of homeless adults (30 percent) in the statewide Minnesota sample were sampled in non-metropolitan areas (i.e., outside of the seven-county metropolitan area that includes Minneapolis and St. Paul Twin Cities) (Wilder Research Center, 2007a, b), and about half of homeless youth (ages 17 or younger) were sampled from non-metropolitan areas (Wilder Research Center, 1998). While this study collected a convenience sample that does not likely reflect the true distribution of homeless adults and youth throughout the state, it documented a large proportion of homeless adults in the non-metropolitan areas (Wilder Research Center, 2007a).

During February and March of 1993, Kentucky conducted a statewide survey of 2,484 adults in both rural and urban service facilities that met the HUD definition of homeless (Kentucky Housing Corporation, 1994). In 2006, a one-night statewide survey of 2,311 homeless persons in Montana was collected (Montana Council on Homelessness, 2007). A broad definition of homelessness was used that included people living doubled up with friends or family or in transitional housing facilities (including domestic violence shelters), foster care, jail, prison, or prerelease settings. Due to the dispersal of the states population, the entire state was treated as rural for the purposes of this report and includes homeless persons from many frontier areas.

The most significant report among the recent literature is the National Survey of Homeless Assistance Providers and Clients (NSHAPC) (Burt et al., 1999), with results stratified by central city, urban and suburban fringe, and rural areas (See Exhibit 1.) This descriptive study was based on samples of individuals drawn from 16 types of homeless service sites, although authors note that because shelters, meal programs, and dedicated homeless services sites are much scarcer in rural areas, homeless adults in rural areas may be underrepresented in their sample. Despite these limitations, the NSHAPC is the most authoritative and most often cited study of rural homeless persons. Not surprisingly, given the variation in their methodologies, locations, and goals, contradictions among findings of extant studies appear. Nevertheless, a profile of rural homeless persons in the United States can be constructed from a review of the existing literature.

Sociodemographic Characteristics

In most state and local studies, the majority of rural homeless adults are single males (Aron, 2004; Center for Family and Community Life, 2005; Montana Council on Homelessness, 2007; Patton, 1988; Vernez et al., 1988). More than three-quarters of the national NSHAPC rural sample were men (77 percent) (Burt et al., 1999). Similarly, males constituted a higher majority of rural homeless adults in California than of urban homeless adults (Vernez et al., 1988). Non-urban homeless adults in the Ohio study were also mostly male, although the non-urban counties had twice as many women as men (Roth et al., 1985; Roth & Bean, 1986). In Montana, virtually equivalent numbers of males and females were surveyed. In contrast, in the statewide survey in Kentucky the majority of rural homeless adults were women (62 percent) whereas urban homeless adults were mostly male (68 percent) (Kentucky Housing Corporation, 1994). In a follow-up to an earlier survey of key informants, which reported a preponderance of homeless men (Vermont Department of Social Welfare, 1985), the number of homeless women in Vermont was estimated to be increasing (Vermont Department of Social Welfare, 1987).

Most studies report that the majority of rural homeless adults are non-Hispanic whites (Aron, 2004; Center for Family and Community Life, 2005; Montana Council on Homelessness, 2007; Patton, 1988; Vernez et al., 1988). For example, homeless adults in the non-urban counties in Ohio were mostly white (Roth et al., 1985; Roth & Bean, 1986) as were homeless adults in rural California (Vernez et al., 1988). On a national level, NSHAPC (Burt et al., 1999) reported that persons without housing in rural areas compared to those in urban areas were more likely to be white. Among homeless youth in non-metropolitan Minnesota, however, racial and ethnic minorities were vastly overrepresented, with 53 percent being African American, American Indian, Hispanic, Asian, or mixed race compared to 18 percent of all Minnesota youth. However, racial and ethnic minorities were even higher among metropolitan homeless youth (80 percent) (Wilder Research Center, 2007a, b). Similarly, in Montana, American Indians were disproportionately represented among homeless persons surveyed, especially among women. American Indians represented 20 percent of all persons identified as homeless, which was 3.2 times higher than reported in the 2000 Montana Census of the general population. Two-thirds of all homeless women were American Indian, while less than 40 percent of homeless men were American Indians. Other minority groups were also overrepresented, constituting 4 percent of persons identified in the 2006 survey compared to 2 percent of the general Montana population (Montana Council on Homelessness, 2007).

Most reports indicate that people who are homeless in rural areas are somewhat younger than those in urban areas. For example, in Ohio, persons sampled in the non-urban counties were slightly younger (ages ranged from 16 to 83) (Roth et al., 1985; Roth & Bean, 1986) than in the urban areas. The estimated average age of homeless people in the Vermont statewide survey was in the early 30s; persons under 18 years of age were estimated to comprise 15 percent of the total, and 8 percent were over 60 (Vermont Department of Social Welfare, 1987). In California, rural homeless adults were more likely to be under age 35 than were those surveyed in urban counties (Vernez et al., 1988). However, in the NSHAPC, rural homeless people were older than homeless adults in urban and suburban areas, with most between the ages of 35 to 44 (Burt et al., 1999).

Nationally, single adults represent the largest portion of the homeless population, and single women with children comprise the great majority of homeless families (Burt et al., 1999). The composition of rural homeless populations is similar: in the NSHAPC national survey, most rural respondents were single adults (84 percent), consisting of 77 percent single men and 7 percent single women (Burt et al., 1999). In rural Vermont, the homeless population was estimated to include 80 percent single adults, the majority of whom were male (70 percent), but with an increasing number of families (Vermont Department of Social Welfare, 1985; 1987). In contrast, in the statewide survey of 2,311 homeless persons in Montana (Montana Council on Homelessness, 2007), all of whom were considered to be rural for the purposes of this report, virtually equal numbers of males and females were identified, with 36 percent of the sample being families with children. Twenty-three percent of the NSHAPC rural sample were women, and most of these (74 percent, or 17 percent of sample) had one or more children with them (Burt et al., 1999). Among non-metro homeless adults in Minnesota, 26 percent were single women with children, 7 percent were couples with children, and 1 percent was single men with children; among metro homeless adults, 23 percent were single women with children, 3 percent were couples with children, and 1 percent was single men with children (Wilder Research Center, 2007a, b). In contrast, in California, none of the homeless adults sampled in non-urban counties had children with them (Vernez et al., 1988). Homeless people surveyed in Ohio non-urban counties were more likely to be married or living with a partner than in urban areas (Roth et al., 1985; Roth & Bean, 1986).

There is some evidence that rural homeless adults are less educated than their urban counterparts. For example, NSHAPC (Burt et al., 1999) reported that persons without housing in rural areas compared to those in urban areas were less likely to have completed high school. However, no differences by educational attainment between rural and urban samples were found in the Ohio study (Roth et al., 1985; Roth & Bean, 1986). In Montana, two-thirds of the rural sample had completed high school or better (66 percent) (Montana Council on Homelessness, 2007).

Studies indicate that rural homeless adults tend to have higher employment rates than their urban peers, but more of them are underemployed. For example, the NSHAPC rural sample was more likely to be working (65 percent)  although often underemployed and in the informal labor market  and less likely to be receiving any means-tested government benefits compared with the urban sample (Burt et al., 1999). Similarly, in the Ohio study, the non-urban sample had more resources and were more likely to be currently employed (35 percent vs. 22 percent urban), but they were less likely to receive welfare benefits or to use meal programs or shelters than the urban sample (Roth et al., 1985; Roth & Bean, 1986). In Montana, 28 percent of homeless adults with a high school education or less were working full- or part-time (Montana Council on Homelessness, 2007).

The experience and trajectory of homelessness among rural settings and rural subgroups is not well documented. For example, non-urban and urban adults in Ohio were equally likely to have been homeless less than one year (median 60-days homeless overall). In Montana, families with children tended to be homeless for shorter periods of time than others in the sample (i.e., 63 percent had been homeless less than 6 months compared to 52 percent of others) (Montana Council on Homelessness, 2007). Only about 5 percent of the sample had been chronically homeless (i.e., for 12 months or longer), and more than half of these had been homeless for more than two years. In the Minnesota study, non-metro homeless adults were less likely to have histories of chronic homelessness (47 percent non-metro vs. 57 percent metro) (Wilder Research Center, 2007a, b).

In Ohio, non-urban respondents were only one-third as likely to have spent the previous night in a shelter as the urban sample; no one in the purely rural counties had stayed in a shelter. By contrast, the non-urban sample was more likely to have stayed with friends or family than the urban sample (Roth et al., 1985; Roth & Bean, 1986). Rural adults in California were more likely than urban adults to have been recruited from the streets than from a shelter (Vernez et al., 1988). In Montana, respondents spent the previous night with family or friends (22 percent), outside (20 percent), in an emergency shelter (18 percent), or in transitional housing such as domestic violence shelters (16 percent) (Montana Council on Homelessness, 2007). One fourth of families with children reported staying in transitional housing facilities. Chronically homeless persons most commonly spent the night outside, in emergency shelters, or in a motel.

In the NSHAPC study, non-urban adults were more likely to have ever been incarcerated than those in urban settings (Burt et al., 1999). In contrast, non-urban adults in Ohio included slightly fewer veterans and ex-offenders (Roth et al., 1985; Roth & Bean, 1986). The Montana study reported that most women in prison in Montana have committed drug-related offenses, usually involving methamphetamine, that range from possession or manufacturing of drugs to committing crimes to get money for drugs (Montana Council on Homelessness, 2007).

Physical Health Problems of Homeless Persons in Rural Areas

People who are homeless in rural areas have greater health problems but less access to health care (Center for Family and Community Life, 2005; National Coalition for the Homeless, 2006). For example, in Ohio, higher rates of health problems were found in the purely rural counties (41 percent versus 20 percent in mixed counties and 41 versus 31 percent in urban counties) (Roth et al., 1985; Roth & Bean, 1986). In Montana, diagnoses of hypertension (11 percent), asthma (11 percent) and hepatitis C (9 percent) were the most commonly reported diseases (Montana Council on Homelessness, 2007). In a series of qualitative interviews on the health of rural homeless persons, Post (2002) asked clinicians nationally to identify the medical conditions that seem to distinguish rural from other homeless clients and to identify obstacles that prevent clients from getting the health care and social support they need. Clinicians described the morbidity from chronic medical conditions such as hypertension or diabetes as being greater than in urban settings because rural clients remain untreated longer than their urban counterparts. Infectious diseases (including hepatitis C, which is associated with injection drug use) were reported to be a growing problem among rural homeless clients, as were sequelae of alcohol and drug problems. Tuberculosis was reported to be more prevalent among recent immigrants from Latin America and Southeast Asia who often work in rural areas. HIV is often diagnosed later in rural areas than in urban areas, after the illness has reached more advanced stages, making treatment more difficult. Traumatic injuries and musculoskeletal disabilities secondary to trauma or injuries from manual labor were also reported (Post, 2002). Several studies identified specific health problems such as HIV/AIDS as causes of homelessness in their samples (Montana Council on Homelessness, 2007; Post, 2002; Wilder Research Center, 2007a, b).

Behavioral Health Problems of Rural Homeless Persons

Sufficient standardized epidemiological studies have not yet been conducted to clarify the incidence, types, patterns, severity, and trajectories of mental health and substance use problems among rural homeless persons. Nevertheless, growing evidence supports the prevalence of behavioral health problems among this population. In the national NSHAPC study, the great majority of homeless adults from all areas (rural, central city, and urban/suburban) reported at least one behavioral health problem (i.e., alcohol, drug, or mental health) in the previous 30 days (6467 percent), during the past year (7272 percent), and in their lifetimes (82 percent87 percent) (Burt et al., 1999). However, significant differences in prevalence of specific disorders were found between rural and urban homeless samples. Rural homeless adults had more current problems with alcohol; however, they had fewer current problems with mental illness or drug use compared to more urban homeless clients (Burt et al., 1999). For example, compared to central city and suburban homeless adults, rural homeless adults consistently reported dramatically higher rates of alcohol use  50 percent of the rural homeless adults reported alcohol problems in the previous 30 days, and 66 percent reported alcohol use problems in their lifetimes. Rural homeless adults also reported generally high rates of mental health, drug use, or comorbid psychiatric and substance use problems; however, the rural rates were dramatically lower than rates reported by the more urban homeless adults.

In the California survey, homeless adults were screened for probable lifetime major mental disorders and lifetime substance use disorders with an instrument derived from the Diagnostic Interview Schedule (DIS-III) (Robins et al., 1981). Because the study used a short screener that identified people who have a high probability of having a lifetime disorder rather than the entire diagnostic interview, the rates of disorder are likely overestimated (Vernez et al., 1988). Nevertheless, homeless persons in the rural county screened higher for severe mental illness (SMI) (especially major affective disorder), substance use disorder (especially for alcohol), and for dual disorders (mental and substance use disorders). The rural county sample (Yolo County) screened positive for severe mental disorders, including high rates of major affective disorders (e.g., major recurring depression and bipolar disorders) and schizophrenia. The rural county had a higher rate of severe mental illness compared to the two more urban counties (Alameda and Orange Counties) (Vernez et al., 1988). Virtually all subjects who screened positive for severe mental disorder also screened positive for either alcohol or drug disorders or both, for a dual diagnosis rate of 37 percent of the total sample (Vernez et al., 1988).

In Montana, self-reported diagnosed conditions included mental illness (18 percent) and alcohol or drug abuse (15 percent) (Montana Council on Homelessness, 2007). About one-quarter of the total sample (26 percent) reported that chronic drug or alcohol abuse was a cause of their homelessness. Among these, 60 percent were male and 40 percent were female. About 17 percent of families with children identified drug or alcohol abuse as a cause of their homelessness.

Because proxy measures for mental health problems (such as psychiatric hospitalization) are often used in studies, the identification of specific mental health problems and their severity is impossible (Patton, 1988; Robertson & Greenblatt, 1992). For instance, in the Ohio study, non-urban and urban samples were about equally likely to have histories of psychiatric hospitalization, and they had similar rates of psychiatric symptoms and similar levels of perceived mental health status and life satisfaction (Roth et al., 1985). Vermont informants estimated that 30 percent of rural homeless adults were deinstitutionalized, that is, people who in years past would have been sent to a state hospital, before changes to the mental health commitment law (Agency Planning Division, 1986). In the Post (2002) study, clinicians serving rural homeless persons reported seeing a number of clients who were disabled with serious mental illnesses (e.g., schizophrenia and affective disorders such as depression or bipolar disorder) and other serious conditions (including personality disorders and posttraumatic stress disorder secondary to childhood abuse, domestic violence, or war-related injuries).

In the general U.S. adult population, substance use and abuse are less common in non-metropolitan areas compared to metropolitan areas, and within non-metropolitan areas, substance use is lowest in the most rural areas (Strong et al., 2005). The prevalence and distribution of substance use problems (alcohol or illicit drugs) reportedly varies by region and population (Post, 2002). In the California survey, the rural county had higher rates of any substance use disorders (alcohol or drugs) compared with the more urban counties. Specific findings on alcohol use and related problems among rural homeless adults are mixed. Some researchers have reported higher rates of alcohol problems among homeless persons in rural areas ( Burt, 1996; New Freedom Commission on Mental Health, 2004; Office of Rural Mental Health Research, 2003; Vernez et al., 1988), while others report lower alcohol problems in rural areas (The Conservation Company, 1989, cited in Housing Assistance Council, 1991). In the NSHAPC national sample, rural clients had more current problems with alcohol compared to more urban clients (Burt et al., 1999). In the Ohio survey, few non-urban respondents identified alcoholism as a problem, although most had used alcohol in the previous 30 days and 20 percent had sought help for alcohol problems in their lifetimes (Roth et al., 1985; Roth & Bean, 1986).

Service providers report serious levels and severity of drug use across many rural areas (Substance Abuse and Mental Health Services Administration, Center for Mental Health Services, 2006), yet empirical quantitative studies are limited so there is minimal assessment of the extent and specific nature of the drug problem among rural populations. For example, in her extensive interviews with health care providers, Post (2002) cites repeated anecdotes about extensive drug involvement of rural homeless persons, with the specific drugs used varying by geographic area. For example, prescription drugs such as OxyContin (also known as synthetic heroin) has become so widely used in places like Appalachia and rural Maine as to be called hillbilly heroin (National Institute on Drug Abuse & Community Epidemiology Work Group, 2001). Other more traditional street drugs, such as cocaine, heroin, or methamphetamine, are reportedly used in diverse rural areas. Anecdotal reports indicate that methamphetamine use in particular is increasing in rural areas (Strong, et al., 2005; Wilder Research Center, 2007a). Reportedly, methamphetamine use began in California and spread eastward as users started manufacturing the drug in home labs in rural areas in the South, Midwest, and other areas (Alligood, 2001, cited in Post, 2002). Empirical evidence of rural/urban differences in drug use is mixed. In the California study, the rural county had a high rate of drug abuse or dependence, which was similar to one urban county but lower than another urban county (Vernez, 1988). In the NSHAPC national sample, rural clients had fewer current problems with drug use compared to more urban clients (Burt et al., 1999). In the Ohio study, non-urban and urban counties reported similar rates of illicit drug use or problems with drugs (16 percent versus 15 percent urban) (Roth et al., 1985; Roth & Bean, 1986).

Access to Health and Behavioral Health Services

Although substantial rates of mental health and substance use problems have been documented among rural homeless persons, efficacy of treatment interventions for these problems in diverse rural populations has not been demonstrated. Many structural, sociocultural, and personal barriers reduce access to rural services generally and to health services in particular (Strong et al., 2005). Although some of these barriers exist in urban settings as well, their impact may be disproportionately greater in resource-poor rural areas, and their impact on homeless and other indigent rural populations is likely to be more severe. Primary structural barriers in rural areas include geographic dispersion and low population densities that increase the cost of services per capita compared to urban areas. Other structural barriers include scarce services, especially for mental health and behavioral health care; inappropriate service models (e.g., urban-based models not adapted to the needs of people living in particular rural areas); lack of outreach to engage rural homeless persons in services; inaccessible health and other services (especially in small rural communities and frontier areas since service programs tend to concentrate in larger rural communities); fragmented systems of health care; lack of cultural competence on the part of program staff; lack of accessible transportation, with greatest impact on families with children and on disabled or older persons; insufficient numbers of health care professionals, particularly specialists; and lack of acceptance by program staff (Patton, 1988; Post, 2002; Strong et al., 2005).

In rural areas, primary care physicians are the main health care providers for persons with diverse health and behavioral health care problems, including comorbid medical, mental health, and substance use problems. These physicians may lack adequate training or experience to treat complex health problems in such a diverse patient population; further, healthcare settings may be understaffed. In many rural settings, recruitment of physicians may be limited to persons providing time-limited services as conditions of training programs, which may present cultural barriers and lack of continuity due to turnover of professional staff.

Sociocultural factors can reduce or delay help-seeking by rural families and individuals, especially for sensitive health and behavioral problems. Many rural communities are characterized by close social ties, reluctance to seek outside assistance, a desire for privacy (especially regarding sensitive problems such as domestic violence or alcohol, drug, or mental health problems), and a tradition of voluntary social support from the community (rather than from fee-based formal agencies) (Strong et al., 2005). Such structural, sociocultural, and personal barriers may result in less use of needed care due to delayed help-seeking. Such delays can lead to more severe symptoms and chronic health problems before the homeless person finally receives care (Post, 2002), which may then entail more intrusive or expensive treatment.

Major personal barriers to care have been reported by health care providers and others, including lack of financial resources and medical insurance; medical disabilities or behavioral health problems that reduce functional status or ability to navigate the available health care system; shame, hostility, or lack of trust, especially for mental health services; a tradition of dependence on self or family and friends; and cultural differences from service providers that may further limit the amount and quality of care available for medical and behavioral health problems (Patton, 1988; Post, 2002; Strong et al., 2005). For example, NSHAPC findings showed that homeless clients in rural areas nationally were less likely to have health insurance (including Medicaid) compared to clients in more urban areas. Furthermore, they were twice as likely as urban homeless clients to have missed getting medical care that they needed in the previous year (Burt et al., 1999). Many in the rural sample reported poor health and having histories of inpatient psychiatric treatment as well as residential treatment for substance use (Vernez et al., 1988). Despite higher apparent need for treatment, however, the rate of any mental health treatment was much lower in the rural county than in the more urban counties, including lifetime psychiatric hospitalization (in Yolo County, 14 percent among people with serious mental illnesses and 0 percent among others). Furthermore, none of the rural county homeless population with SMI had been hospitalized in the previous 12 months and few had received outpatient treatment (6 percent of SMI) or psychiatric medications (9 percent of SMI) in the previous 6 months (Vernez et al., 1988). In contrast, in the Ohio study, there were only minimal differences by county type in health status and utilization of emergency treatment and other health services, and of mental health problems (Roth et al., 1985; Roth & Bean, 1986). However, the urban adults were more likely to have been discharged with no arrangements for follow-up care (37 percent) compared to the non-urban group (24 percent) (Roth et al., 1985).

Reasons for Homelessness

While systemic factors, including lack of infrastructure (e.g., child care and public transportation) to support employment; long distances between low-cost housing and employment opportunities; and inadequate treatment opportunities for disabling medical and behavioral health problems, including serious mental health and substance use problems, are frequently cited as causes of rural homelessness (Center for Family and Community Life, 2005; National Coalition for the Homeless, 2006b), research studies report personal reasons as contributing to individuals homelessness. For example, in the Vermont survey, key informants reported that homelessness was related to poverty, reduced housing options, and mental health and substance abuse problems. This contrasts with results of a follow-up study conducted in 1987, (Vermont Department of Social Welfare, 1987), which related homelessness to federal cuts in housing, welfare, and services; high housing costs; low-paying jobs for unskilled workers; deinstitutionalization; and the increase in single-parent households. In addition, studies report domestic violence as a major contributor to homelessness of women (Kentucky Housing Corporation, 1994; Intergovernmental Human Services Bureau, 2003) and unaccompanied youth (Wilder Research Center, 1998). For example, in Ohio, although about half of both groups reported economic reasons for not having a home, the non-urban group reported more family-related reasons, with non-urban women less likely to have used domestic violence shelters (Roth et al., 1985; Roth & Bean, 1986). In addition, on occasion natural disasters (such as Hurricane Katrina) contribute to homelessness in rural areas through displacement of formerly housed persons.

Effective Service Models for Rural Homelessness

The research literature addressing rural homeless service models is very limited. Little research has evaluated rural programs, compared them to urban programs, or evaluated homeless service strategies unique to rural areas. Moreover, the available program evaluation and research tends to focus narrowly on one type of program, one particular program, or one location. Although valuable information concerning promising practices can be gleaned from providers and advocates, these service models have not been subjected to rigorous testing and have not been widely disseminated. An examination of state, regional, and city plans to end homelessness, completed in late 2006, indicates that only 1 percent of the 90 plans examined focuses on strategies to end homelessness in rural areas (National Alliance to End Homelessness, 2006). However, some states updated their plans since the NAEH study with the result that specific goals and strategies related to rural homelessness are beginning to emerge. The Montana plan, for example, calls for partnerships with tribes to learn more about homelessness in rural or reservation settings. South Dakota plans include capacity expansion in rural areas, including reservations, with a goal of developing housing and service models for rural areas. In response to a biannual study of homelessness in 2005, Iowa has revised its plan to end homelessness to include strategies for providing equitable access to health services to rural residents (Ditsler et al., 2007; Montana Council on Homelessness, 2007). In addition, 17 rural and Appalachian counties in southeastern Ohio have joined together to develop and implement plans to end homelessness. In 2007, the project, Rural Homeless Initiatives in Southeast and Central Ohio, will produce the first regional, county-driven plan to end homelessness in a rural area.

Two studies currently testing service models for homeless populations in both urban and rural areas include extensive evaluations. The Supportive Housing and Managed Care Pilot in Minnesota is a permanent supportive housing project operating since 2001 in Ramsey County (an urban county including St. Paul and its suburbs) and Blue Earth County (a rural county including Mankato and the surrounding area). Blue Earth County has a population of approximately 58,000 residing in an area of 764 square miles, including the small town of Mankato (population 32,427). Early findings show that the families and single adults in Blue Earth County tend to enter the pilot with a higher level of functioning than their urban counterparts but have the same trajectory of progress (Jennifer Ho, personal communication, 2007). Trusting relationships with staff appear to hold promise for the development of long-term stability in the community (National Center on Family Homelessness, 2006). Stakeholder interviews indicate that individuals and families with long histories of homelessness, mental illness, substance use disorders, and HIV/AIDS are achieving housing stability. A cost study and an adult outcome study are scheduled for release in 2007.

The Sound Families Initiative, established by the Bill and Melinda Gates Foundation, has as its goal to create 1,500 service-enriched transitional housing units for homeless families in three counties in Washington, two urban (King CountySeattle and Pierce CountyTacoma) and one rural (Snohomish County), and to serve as a catalyst for a new level of cooperation on homelessness-related issues within the three counties. The mixed-method evaluation examines the impact of Sound Families on the families served, on transitional housing programs, as well as on the challenges the programs and their clients continue to face. The evaluation has nearly 200 families enrolled in a longitudinal interview process at 10 case study sites. Although the first reports do not compare the families living in the rural county with those in the two urban counties, families reported improvement in their overall quality of life, with 88 percent securing permanent housing at exit and 74 percent increasing their household income. In addition, the number of families with children who attended only one school during a school year increased from 49 percent at intake to 80 percent at exit and 86 percent one year after exit. Attendance improved for the children in stable housing, and parents reported their children were doing better in their schoolwork (Bodonyi et al., 2005).

Service Strategies

In 1999, Goodfellow began looking at the differences in rural and urban homeless service providers (Goodfellow, 1999). She discovered that there may be more differences among homeless service delivery systems in rural and urban areas than among the persons experiencing homelessness in the two areas. Through interviews with clinicians, Post (2002) identified rural service models that vary in size and distance from urbanized areas. In larger areas, strategies include community partnerships that link formal and informal support systems, multi-service centers, and hub-and-spoke models of outreach to and referrals from outlying rural and urban communities. In smaller rural or frontier areas, where there are only minimal services for people who are homeless, clinicians have two strategies: mobile outreach units and, as a last resort, Greyhound referrals or Greyhound therapy, that is, transport to urban areas that have the needed services (Post, 2002).

Clinicians have recommended strategies to overcome barriers to health care and to prevent rural homelessness. These include:

• integrating behavioral health care with primary care services to reduce or eliminate the powerful social stigma associated with mental illness in rural areas;
• providing transportation assistance to address the scarcity of public transportation in rural areas;
• expanding health coverage and facilitating access to covered services;
• expanding health care entitlement programs to cover low-income persons who are homeless or at risk of homelessness;
• developing a service delivery infrastructure in rural communities responsive to needs of people who are homeless (including temporary shelter services and basic health and social services);
• coordinating rural service delivery systems to maintain continuity of care;
• increasing outreach to hidden homeless people in remote rural areas;
• using community networks and indigenous workers to facilitate mobile outreach;
• promoting cultural competence of service staff to address communication and other cultural barriers;
• conducting early interventions for youth, families with children, and single adults who are newly homeless or at risk for homelessness; and
• focusing on prevention of homelessness by addressing structural causes of poverty (Post, 2002).

Best and Promising Practices

The literature on best practice models for services for people who are homeless focuses primarily on urban models, providing no comparative findings on the use of the same models in rural communities or on homeless service program models developed specifically for rural areas. Two reports produced by the HAC address related best practices for rural areas. Continua of Care Best Practices: Comprehensive Homeless Planning in Rural America, provides case studies of four rural communities that have successfully created and maintained rural systems and homeless shelter and service projects (Housing Assistance Council, 2002). Data were collected from site visits, interviews, and examination of the 1999 continuum of care applications. Based on analyses of these data, researchers identified four common themes across these successful programs: leadership, inclusive process, planning, and support networks.

Best Practices in Revolving Loan Funds for Rural Affordable Housing (Housing Assistance Council, 2003a), analyzes four case studies of rural revolving loan funds, identifies similarities and differences, and provides specific advice for those seeking to establish such funds. Researchers sought to ascertain which best practices are most salient in different rural contexts. The study recognized that what works best in one rural community may not be appropriate in another, thus calling for careful examination of the impact of the social and economic context of community lending practices.

Despite scant research, there is some evidence pointing to models emerging as promising practices. These include regionalized services, development of community collaboration and coalitions, rural service teams, the housing-plus-services model, and employment initiatives. For example, a HUD continuum of care initiated by 50 service providers in a 23-county rural region of western Tennessee in 2004 brings in $1 million annually for services that help support 600 units of housing built using a combination of HUD and private funding (Rozann Downing, personal communication, 2007).

One of the greatest assets in rural communities is the ability  or necessity  to collaborate in the provision of housing and services due to limited resources. In fact, the need to provide services to clients with complex needs drives the organization to create interconnections to a set of other service providers (Goodfellow, 1999). This decreases cost, increases community building, and reduces duplication in service delivery. While limited resources in rural communities mean that proportionately more resources and funding sources are needed to create programs than in urban communities, once this is accomplished, the rural programs are more stable due to diversity of funding and other resources (Housing Assistance Council, 2006). This necessity also resulted in rural communities being the first to make full use of mainstream resources to serve homeless people (Burt, 1996). The work of Hazard Perry County Community Ministries in Southeastern Kentucky, in the Appalachian foothills, exemplifies some of the unique strengths found in rural communities attempting to address the issue of homelessness. The success of this agencys program models is attributed to using lay workers who know the community and understand the people they are serving, being creative in interpreting the regulations of mainstream programs so that people in need can be served, and maintaining an organizational culture that doesnt give up on people (Gerry Roll and Jennifer Weeber, personal communication, 2006). Rural citizens are more likely to have lifelong relationships with their service providers and other members of the community, creating strong community bonds, trust in service providers, and the desire to help one another. (Burt, 1996). This makes it possible to address the needs of persons in need more quickly and increases the likelihood of long-term success (Housing Assistance Council, 2006).

Rural service teams from across various agencies and organizations support the families and single adults in the previously mentioned Supportive Housing and Managed Care Pilot in Blue Earth County, Minnesota. A challenge faced by rural teams is having fewer disciplines represented, because teams are smaller on smaller-scale projects. This also means that cost efficiencies are fewer. An advantage for rural teams, however, is less bureaucracy to get in the way of linking people to the community services and supports they need (Jennifer Ho, personal communication, 2007). The pilot model, which demonstrated success in both rural and urban areas, was expanded in summer 2006 into the seven counties of rural northeastern Minnesota, including three Indian reservations, as well as into some southern Minnesota counties and the metropolitan area of MinneapolisSt. Paul.

Both the Tennessee and Minnesota initiatives described here are permanent supported housing models. Support services, whether offered by providers located on site or off site, are a critical component of the response to homelessness in both rural and urban areas. The smaller population in rural areas usually means fewer units in a project, with projects farther apart than their urban counterparts. The Millennium Center in Cuthbert, Georgia, consisting of 20 freestanding residences for homeless and near-homeless families with alcohol and/or drug addiction issues, offers a full spectrum of supports. There is a child care center and a satellite of Albany Technical College on site as well as treatment services for the whole family. About 60 percent of the 80 families who have lived at the Center since it opened in 2003 have adult family members who have gotten sober, secured educational services and a job, and moved on to permanent housing. Cuthbert is in the heart of a seven-county area of very rural Georgia with a total population of 38,000.

Employment demonstration projects in rural areas of Illinois, Nebraska, and Tennessee offer findings and lessons learned similar to other efforts that target rural poverty. Lack of transportation makes it difficult for low-income participants to access jobs and other services. Jobs are scarce and low paying, often providing less than full-time work. Partnering among local organizations is critical to client success, as is having program staff members who are familiar with the rural communities they serve and who can work independently. Outreach services are essential in covering rural areas with large territories (Burwick, Jethwani, & Meckstroth, 2004). Employment opportunities are available to participants in a unique transitional living program operated by Good Works, Inc., in the Appalachian area of southern Ohio. Most residents are offered work in two agency-owned businesses, a gift shop and a bed and breakfast. Residents are linked with volunteer live-in mentors who are available for personal support, modeling the development of healthy relationships, and helping residents develop and carry out plans for future stability (Wasserman, 2006).

Directions for New Research

Gaps in Current Research

The literature that constitutes this first generation of research on rural homelessness has broken difficult ground and laid a foundation for the next generation of studies. Nevertheless, limitations of this generations empirical work include problems with design, sampling, and instrumentation. Serious limitations of many studies in the existing literature include sampling from limited service sites; using idiosyncratic definitions of homelessness; collecting minimal information on smaller or more isolated rural or frontier areas; failing to collect data on household composition, which affects demand on social service systems; and lacking standardized instrumentation that would facilitate comparisons across studies and with normative populations.

Many of these studies relied on convenience samples collected from clients of homeless services in rural areas. These samples lacked comprehensive coverage of the target population by failing to include homeless persons who do not use services or who live in smaller or more remote rural or frontier areas where services are not available. The lack of comparison groups makes it difficult to identify factors that differentiate the rural homeless population from the rural low-income housed population. Simplistic divisions of samples into dichotomous rural/urban or trichotomous rural/urban/suburban categories potentially leave undiscovered important differences among these groups. Moreover, researchers often combined data from rural and mixed areas into a single category, potentially masking important differences by location and proximity to urban areas. Broad conclusions about rural homeless populations cannot be made without representative data (Strong et al., 2005).

In most studies, instrumentation did not adopt standardized questions about health status. Even the important study by Burt and colleagues (1999) used non-standardized assessments to measure alcohol, drug, and mental health problems. The lack of a core of common questions makes comparisons across studies difficult. There was minimal assessment of drug use in studies reviewed, despite reports by providers that drug use is an increasingly urgent concern for rural communities. The lack of longitudinal designs leaves many questions unanswered about the patterns of homelessness experienced by these rural populations and their utilization of services over time.

The limited research available leaves many questions unanswered about the rural homeless population. As was true in initial studies of urban homeless populations, variations in research methodologies, particularly sampling and instrumentation, provided contradictory findings concerning important population characteristics. Given the current state of the research on rural homelessness, it is not possible to determine with certainty whether rural homelessness is distinct from urban homelessness in terms of sociodemographic or behavioral health profiles. For example, some findings have suggested that persons who are homeless in rural areas are more likely to be families, employed, and better educated than their urban counterparts. Studies as well as providers indicate that persons who are homeless in rural areas are more likely to be found in some type of substandard housing rather than to be literally homeless as is seen in urban settings. Nevertheless, whether or not future research identifies distinguishing characteristics between rural and urban homeless populations, it is abundantly apparent that the differences between rural and urban services infrastructures alone mandate that further research on the needs of rural homeless populations be pursued.

Methodological Challenges for Research with Rural Homeless Populations

More studies are needed that include rural areas, rural systems, and rural populations. Future research on people who are homeless in rural areas can build on the seminal first generation of work. The most relevant and rigorous methodologies developed for study of urban homeless populations and service systems, such as representative sampling, longitudinal designs, standardized instrumentation, and mixed qualitative and quantitative methodologies, can be adapted to explore the rural population that is still hidden to researchers. Some specific methodological challenges are outlined below.

Lack of common definitions. Development of a definition of the rural/urban continuum that captures the diversity across population densities and geographic locations is necessary for comparing findings across studies. Since rural populations in the United States are not homogenous, study results are often determined by how rural and homelessness are defined (National Institute on Drug Abuse, 1997). Whatever definitions are used must be carefully operationalized for each study and described to permit replication and comparison of findings (New Freedom Commission on Mental Health Subcommittee on Rural Issues, 2004; Rural Information Center, 2006; Strong et al., 2005). In addition, rural/urban typologies used in recent research on rural homelessness have been too crude to capture many true differences between rural and urban communities and between various types of rural communities. Given the apparent variation in medical and behavioral health problems by size and locations of rural communities, it will be necessary to study the epidemiology of health, mental health, and substance use disorders as a function of population density and geographic placement on a continuum rather than studying only dichotomous urban/rural or trichotomous urban/rural/suburban comparisons (National Institute on Drug Abuse & Community Epidemiology Work Group., 2001; Patton, 1988).

Sampling strategies. The most representative sampling strategy is desirable but often not considered feasible. A sampling strategy based on service sites must use a broad range of services across the range of rural areas (and not rely only on dedicated homeless services since they are frequently absent in smaller rural or frontier areas). Such sites might include, for example, welfare and other social service agencies, public and mental health departments, free clinics, employment centers, and other broad-based service programs (Aron, 2004). Many of the existing studies use such small numbers of rural subjects that characteristics of rural populations are masked by the usually larger urban sample. Also, variations in conditions between urban and rural populations or between varied types and sizes of rural communities are not possible to explore (e.g., differences in drugs used, psychiatric conditions experienced, or patterns of alcohol use) when insufficient rural sample sizes are used. Some aspects of research on rural communities may have limited generalizability, because rural communities exhibit unique regional character; for example, Appalachia and the western frontier have distinctive cultures that may affect homelessness through deeply held values such as importance of family structure, insularity, independence, and so forth. It is difficult to do cross-cutting research because of differences in the character of a given rural area compared to other types of rural areas, and compared to non-rural areas. Conducting multisite studies may clarify aspects of homelessness in rural settings that are site-specific.

Standardized instrumentation for needs assessment and outcomes assessment. Epidemiological studies with standardized instrumentation are needed to clarify the prevalence and severity of medical, mental health, and substance use problems among rural homeless persons. Comparisons across studies for rural populations and comparisons with other groups will be facilitated by appropriate standardized instruments on health status, mental health and substance use disorders, and other domains (e.g., service outcomes and family violence). Also, use of common designations of demographic characteristics (e.g., racial or ethnic group using Census categories) will further facilitate comparisons across studies and with normative populations.

Data sources. A primary difficulty in conducting rural research is finding suitable data sources. It is rare that homelessness is measured in national, regional, or even local surveys and, rarer still, for data to be classified as rural and urban. However, related data sources on other aspects of rural research may indirectly illuminate rural homelessness. Rural Research Needs and Data Sources for Selected Human Services Topics (Strong et al., 2005) provides a thorough examination of data sources available for three human service focal topics. The detailed descriptions of these data sources include an assessment of their strengths and weaknesses for rural research as well as information on their availability and cost to researchers (Strong et al., 2005). For example, data collected as part of the required point-in-time counts for HUD continuum of care applications and other data reports required by various federal grants may provide a source of comparable information on rural homeless populations. However, it is important to be attentive to the quality of the methods used to collect such data, which varies broadly across sites. Data on children who are homeless are collected in every locality by the school board or other entity responsible for funding received through the No Child Left Behind Act of 2001 (U.S. Department of Education, 2004), which requires jurisdictions to locate and enroll homeless children. While the Act endorses the collection of longitudinal data by states and school districts, those entities are not required to incorporate collected data into a longitudinal database. Nevertheless, datasets such as these may be useful for researchers examining rural homelessness.

Recommendations for New Research

There are a number of gaps in current research on people who are homeless in rural areas. Gaps with regard to rural homelessness research are only one aspect of a much larger research gap with regard to rural human services knowledge. In the absence of current empirical studies of rural human services conditions, needs, and programs, policymakers must either ignore rural differences or make assumptions about them. For all these reasons, rural human services research deserves a high priority (Strong et al., 2005). Members of the Expert Panel on Rural Homelessness and others interviewed strongly agreed that rural homelessness research, informed by both available data and current work in the field, must be supported and undertaken in order to inform policy (Substance Abuse and Mental Health Services Administration, Center for Mental Health Services, 2006). The following study areas are recommended.

Special populations.Persons without housing in rural areas are not homogeneous populations  their diversity cannot be overemphasized (Patton, 1988). Moreover, parts of the country, such as Appalachia, the Midwest, the western frontier, and the rural South, have distinctive cultures or regional characteristics informed by local factors such as geography, history, and economy that may have an impact upon the prevalence and trajectory of homelessness (Housing Assistance Council, 1991). Differences between rural and urban populations are often masked by the characteristics of larger urban samples. In order to focus on rural communities, an overrepresentation of rural subjects and rural areas in future homelessness research may be necessary. More research is required to better document the special subpopulations, their sizes, distinctive characteristics, specific service or housing needs, cultural differences, and geographic characteristics of the areas that affect service provision. Both community size and proximity to urban areas have a profound impact on a communitys ability to develop and maintain a formal social services network for the areas residents, including its homeless persons (Patton, 1988). Research is needed to identify the needs of these diverse populations and to develop effective housing and service interventions. However, the existing research is sparse and uneven. Specific next steps in research should include a staged series of studies on homelessness and near homelessness in rural and frontier environments (Substance Abuse and Mental Health Services Administration, Center for Mental Health Services, 2006). The initial stage would include formative qualitative research to lay the foundation for subsequent research, including interviews with frontline service providers (social service program staff, housing authorities, social workers, treatment providers, outreach workers, school staff, and others living and working in rural and frontier areas). Additional efforts would include key-informant interviews and focus groups with people who are homeless or at risk for homelessness. This initial stage would lead to a second stage of carefully crafted survey research, including needs assessments and access to services for persons who are homeless (or at risk) and systematic assessment of systems of service programs. These two stages would inform the third stage, which is development of specific intervention strategies that would include program evaluations. The fourth stage would be dissemination of effective practices to other areas and evaluation of effectiveness in the new sites.

Longitudinal studies of rural homelessness. Rigorous studies of the course of homelessness can help identify precipitating events for a homeless episode in rural communities and pathways out of homelessness for an array of community types across the urban/rural continuum. More rigorous longitudinal research is needed on medical and behavioral health problems among rural homeless persons. Such research should include variations among different types of rural communities, between rural and frontier communities, and between homeless and non-homeless persons in rural areas.

Discharge planning and homelessness prevention. People in rural areas are at risk for homelessness (as in many urban areas) when discharged from public institutions including public hospitals, respite care, psychiatric wards, board and care homes, correctional facilities, foster placements, and other settings (Patton, 1988). Development and evaluation of strategies for discharge planning should avoid missed opportunities to make critical interventions that would follow the persons into the community to help support them against homelessness.

Services research. Research suggests that homeless persons in rural areas are underserved compared to homeless populations in more urban areas. Studies are needed that assess and monitor the availability, accessibility, quality, and outcomes of medical, mental health, alcohol, and drug abuse services for homeless and other indigent individuals in rural areas. Resources include information compiled from nontraditional sources to expedite compilation of good working models or practices from, for example, expert panels, regional meetings, and provider networks. Longitudinal research with large representative samples from across the rural/urban continuum is needed to measure the larger demand for and access to needed services in rural communities (whether broad or homeless-specific services) with consideration for stigma, confidentiality, perceived availability, true availability, and cultural sensitivity (Office of Rural Mental Health Research, 2003). Success for homeless persons is expected to be greater if people can be served early and in their own communities. Research is needed to determine how to make the best use of limited service funds  whether to serve people in their own smaller or remote communities or to centralize services in larger communities. Some services may best be provided locally and others in centralized settings; if so, it is important to distinguish these services. Rural communities are becoming increasingly diverse, and research is needed to identify the impact of changes in population size and composition on the demand for services and housing in rural communities. Research is needed to help rural communities anticipate and plan for needed changes in the local service delivery systems (Office of Rural Mental Health Research, 2003).

Identify best practices for rural settings. There is a paucity of research about best practices for meeting the needs of homeless persons in rural areas, whether adults, families with children, or unaccompanied youth. Research is needed on the effectiveness of case management, supportive housing models, or other models for housing and service delivery for the array of rural and frontier populations. As outlined by the New Freedom Commission on Mental Healths Subcommittee on Rural Issues (2004), most rural advocates believe that because of scarce resources, specialized services are not practical and that service providers are obliged to provide the full range of services to the full spectrum of persons in need. Research is needed to determine which is more cost-effective: addressing homeless persons as a special separate population or together with the low-income housed rural population. Communities that have adequate infrastructure can test the relative efficacy of adapting local mainstream programs to serve homeless clients versus creating homeless-specific services. Studies are needed to assess access to care for both general and specialized health needs. Adaptation of evidence-based practices implemented in urban areas as well as development of models specifically designed for rural areas should be pursued. Developing best practice models should include examination of workforce issues needed to implement model programs and testing the cost-effectiveness of service and housing programs in rural areas generally, including remote rural and frontier communities.

Implications for Preventing and Ending Rural Homelessness

Once evidenced-based practices are established for rural areas, it is important to find ways to disseminate this information to rural communities. Generally, rural communities have much less access to resources for disseminating successful approaches or training providers on their implementation (Office of Rural Mental Health Research, 2003). This is part of what one rural expert called the cycle of disparity (Substance Abuse and Mental Health Services Administration, Center for Mental Health Services, 2006). The lack of resources begins with funding. While an estimated 9 percent of the homeless population live in rural communities, only 5 percent of targeted homelessness assistance reaches these areas (Strong et al., 2005). A significant disparity in federal spending on rural community development has also been seen in recent years: two to five times more per capita is spent in urban areas versus rural (Johnson & Rathge, 2006). In addition, as described in Evaluation of Continuums of Care for Homeless People (Burt et al., 2002), a great deal of complexity is often added to service delivery in rural communities, because funds are distributed on a regional basis, requiring rural service providers to negotiate, plan, and compete for resources on a regional or statewide level. A prime example is the distribution of HUD continuum-of-care funding for persons who are homeless (Burt et al., 2002). While the continuum of care requires all applicants to compete for funding, other dollars for the provision of housing and services often are awarded directly to cities, when rural communities must compete at the state level for the same funds. Examples of this disparity include the Community Development Block Grant and the HOME Program.

The relative lack of funding going to rural communities also means that less is known about what is and is not working. The primary data sources for homelessness research are databases and reports made available to federal funding sources. When these funds do not reach rural communities, the information is not collected and analyzed and program efficacy remains unevaluated. The lack of data and, therefore, research means that public policy responses continue to focus on urban poverty without regard to unique rural factors, including higher poverty rates among children and lack of transportation and specialized services (OHare et al., 2004). Only with a conscious effort to break this cycle can rural homelessness not only begin to be understood, but finally be addressed.

References

Agency Planning Division. (1986). Homelessness in Vermont. Montpelier, VT: Department of Human Services.

Aron, L. Y. (2004). Homelessness, rural. In D. Levinson (Ed.), Encyclopedia of homelessness (pp. 250255). Thousand Oaks, CA: Sage Publications.

Aron, L. Y., & Fitchen, J. M. (1996). Rural homelessness: A synopsis. In J. Baumohl (Ed.), Homelessness in America (pp. 8185). Phoenix, AZ: Oryx.

Bodonyi, J., Orlando, J., Yancey, B., & Erwin-Stewart, N. (2005). Evaluation of the Sound Families Initiative: A closer look at homeless families' lives during and after supportive transitional housing. Seattle, WA: University of Washington School of Social Work, Northwest Institute for Children and Families.

Building Officials and Code Administrators International. (1996). The BOCA national building code model building regulations for the protection of public health, safety and welfare. Country Club Hills, IL: Building Officials & Code Administrators International.

Burt, M. R. & Cohen, B. E. (1989). America's homeless: Numbers, characteristics, and programs that serve them. Washington, DC: The Urban Institute.

Burt, M. R. (1996). Rural homelessness: Focusing on the needs of the rural homeless: What we've learned, what we need to do next. A report on the findings and implications of RECD's rural homelessness conferences. Washington, DC: U.S. Department of Agriculture, Rural Economic and Community Development.

Burt, M. R., Aron, L. Y., Lee, E., & Valente, J. (2001). Helping America's homeless: Emergency shelter or affordable housing? Washington, DC: The Urban Institute.

Burt, M. R., Laudan, Y. A., Douglas, T., Valente, J., Lee, E., & Iwen, B. (1999). Homelessness: Programs and the people they serve. Findings of the National Survey of Homeless Assistance Providers and Clients: Technical report. Washington, DC: Interagency Council on the Homeless.

Burt, M. R., Pollack, D., Sosland, A., Mikelson, K. S., Drapa, E., Greenwalt, K., et al. (2002). Evaluation of continuums of care for homeless people: Final report. Washington, DC: U.S. Department of Housing and Urban Development, Office of Policy Development and Research.

Burwick, A., Jethwani, V., & Meckstroth, A. (2004). Implementing welfare-to-work programs in rural places: Lessons from the Rural Welfare-to-Work Strategies Demonstration evaluation. Final report. Princeton, NJ: Mathematica Policy Research.

Center for Family and Community Life. (2005). Cecil County homeless study 20042005: A survey and analysis of homelessness in Cecil County. Salisbury, MD: Salisbury University.

Child Welfare League of America, National Association for the Education of Homeless Children and Youth, National Health Care for the Homeless Council, National Policy and Advocacy Council on Homelessness, & Volunteers of America. (2005). A matter of definition: Responding to homelessness among families, children, and youth. Washington, DC: National Policy and Advocacy Council on Homelessness.

Ciarlo, J. A., Wackwitz, J. H., Wagenfeld, M. O., & Mohatt, D. F. (1996). Focusing on "frontier": Isolated rural America (Letter to the field No. 2). Boulder, CO: Frontier Mental Health Services Resource Network.

Coburn, A. F., MacKinney, A. C., McBride, T. D., Mueller, K. J., Slifkin, R. T., & Wakefield, M. K. (2007). Choosing rural definitions: Implications for health policy (Issue Brief #2). Columbia, MO: RUPRI Rural Policy Research Institute Health Panel.

Consolidated Plan Revisions and Updates for Department of Housing and Urban Development, 71 Fed. Reg. 6950 (Feb. 9, 2006) (to be codified at 24 C.F.R. pt. 91 and 570).

Cowan, C., Breakey, W.R., & Fischer, P.J. (1988). The methodology of counting the homeless. In Institute of Medicine (Ed.), Homelessness, health and human needs. Washington, D.C.: National Academy Press.

Ditsler, E., Fisher, P., MacDonald, H., & Walz, S. (2007). 2005 Iowa homeless study. Iowa Council on Homelessness.

Fosburg, L. B., & Dennis, D. L. (Eds.). (1999). Practical lessons: The 1998 National Symposium on Homelessness Research. Washington, DC: U.S. Department of Housing and Urban Development and U.S. Department of Health and Human Services.

Goodfellow, M. (1999). Rural homeless shelters: A comparative analysis. Journal of Social Distress and the Homeless, 8, 2135.

Hopper, K., & Hamburg, J. (1986). The making of America's homeless: From Skid Row to new poor, 19451984. In R. G. Bratt, C. Hartman, & A. Myerson (Eds.), Critical Perspectives on Housing. Philadelphia: Temple University Press.

Housing Assistance Council. (1991). Rural homelessness: A review of the literature. Washington, DC: Author.

Housing Assistance Council. (2002). Continua of care best practices: Comprehensive homeless planning in rural America. Washington, DC: Author.

Housing Assistance Council. (2003a). Best practices in revolving loan funds for rural affordable housing. Washington, DC: Author.

Housing Assistance Council. (2003b). Rental housing in rural America. (HAC information sheet). Washington, DC: Author.

Housing Assistance Council. (2006). Formulas for success: Housing plus services in rural America. Washington, DC: Author.

Housing Assistance Council. (2007). From bad to worse: Rural housing in the Administration's FY 2008 budget. Washington, DC: Author.

Hudson, C. G. (1998). Estimating homeless populations through structural equation modeling. Journal of Sociology and Social Welfare, 25, 136154.

Institute of Medicine Committee on Health Care for Homeless People. (1988). Homelessness, health, and human needs. Washington, DC: National Academy Press.

Intergovernmental Human Services Bureau. (2003). Homeless in Montana: A report. Helena, MT: Montana Department of Public Health and Human Services.

Jensen, L. (2006). At the razor's edge: Building hope for America's rural poor. Rural Realities, 1, 18.

Johnson, K. (2006). Demographic trends in rural and small town America. Durham, NH: University of New Hampshire, Carsey Institute.

Johnson, K. M., & Rathge, R. W. (2006). Does rural Great Plains depopulation reflect failed public policy? Ames, IA: Iowa State University, North Central Regional Center for Rural Development.

Kentucky Housing Corporation. (1994). 1993 Kentucky homeless survey preliminary findings. Frankfort, KY: Author.

Kentucky Housing Corporation. (2002). 2001 Kentucky housing survey. Frankfort, KY: Author.

Lawrence, M. (1995). Rural homelessness: A geography without a geography. Journal of Rural Studies, 11, 297307.

Montana Council on Homelessness. (2007). No longer homeless in Montana: A report on the state of homelessness and a ten year plan to end it. Helena, MT: Montana Department of Public Health & Human Services.

Mosley, J. M., & Miller, K. K. (2004, March). What the research says about spatial variations in factors affecting poverty (RPRC Research Brief 2004-1). Columbia, MO: RUPRI Rural Poverty Research Center.

National Alliance to End Homelessness. (2006). A new vision: What is in community plans to end homelessness. Washington, DC: Author.

National Alliance to End Homelessness. (2007). Homelessness counts. Washington, DC: Author.

National Center on Family Homelessness. (2006). The Minnesota Supportive Housing and Managed Care Pilot: Qualitative evaluation, year four. Newton Centre, MA: Hearth Connection.

National Coalition for the Homeless. (2006a). Homeless youth (Rep. No. NCH Fact Sheet #13). Washington, DC: Author.

National Coalition for the Homeless. (2006b). Rural homelessness (Rep. No. NCH Fact Sheet #11). Washington, DC: Author.

National Institute on Drug Abuse. (1997). Rural substance abuse: State of knowledge and issues. Rockville, MD: U.S. Department of Health and Human Services, National Institutes of Health.

National Institute on Drug Abuse, & Community Epidemiology Work Group. (2001). Epidemiologic trends in drug abuse: Vol. II. Proceedings of the Community Epidemiology Work Group. Bethesda, MD: National Institutes of Health, Division of Epidemiology, Services and Prevention Research.

National Rural Housing Coalition. (2004). Preserving rural Americas affordable rental housing: Current issues. Washington, DC: Author.

New Freedom Commission on Mental Health. (2004). Subcommittee on Rural Issues: Background paper. Rockville, MD: Substance Abuse and Mental Health Services Administration.

New Freedom Commission on Mental Health Subcommittee on Rural Issues. (2004). Subcommittee on Rural Issues: Background paper. Rockville, MD: Substance Abuse and Mental Health Services Administration.

O'Hare, W. P., & Johnson, K. M. (2004). Child poverty in rural America. Washington, DC: Population Reference Bureau.

Office of Rural Mental Health Research. (2003). A rural mental health research agenda: Building on success by planning for the future. Rockville, MD: National Institute of Mental Health.

Patton, L. T. (1988). Appendix C: The rural homeless. In Institute of Medicine Committee on Health Care for Homeless People (Ed.), Homelessness, health, and human needs (pp. 183217). Washington, DC: National Academy Press.

Popper, F. J. (1986, Autumn). The strange case of the contemporary American frontier. Yale Review, 101121.

Post, P. A. (2002). Hard to reach: Rural homelessness & health care. Nashville, TN: National Health Care for the Homeless Council.

Ringwalt, C. L., Greene, J. M., Robertson, M., & McPheeters, M. (1998). The prevalence of homelessness among adolescents in the United States. American Journal of Public Health, 88, 13251329.

Robertson, M. J., & Greenblatt, M. (1992). Homelessness: A national perspective. New York: Plenum.

Robertson, M. J., & Toro, P. A. (1999). Homeless youth: Research, intervention, and policy. In L. B. Fosburg & D. L. Dennis (Eds.), Practical lessons: The 1998 National Symposium on Homelessness Research. Washington, DC: U.S. Department of Housing and Urban Development and U.S. Department of Health and Human Services.

Robins, L. N., Helzer, J. E., Croughan, J., & Ratcliff, S. K. (1981). National Institute of Mental Health Diagnostic Interview Schedule. Archives of General Psychiatry, 38, 38189.

Rosenheck, R., Bassuk, E., & Salomon, A. (1999). Special populations of homeless Americans. In L. B. Fosburg & D. L. Dennis (Eds.), Practical lessons: The 1998 National Symposium on Homelessness Research. Washington, DC: U.S. Department of Housing and Urban Development and U.S. Department of Health and Human Services.

Roth, D. & Bean, G. J. (1986). New perspectives on homelessness: Findings from a statewide epidemiological study. Journal of Hospital and Community Psychiatry, 37, 712719.

Roth, D., Bean, J., Lust, N., & Saveanu, T. (1985). Homelessness in Ohio: A study of people in need. Columbus, OH: Ohio Department of Mental Health.

Rural Information Center. (2006). What is rural? Web-based resources. Washington, DC: U.S. Department of Agriculture, National Agricultural Library.

Saulny, S. (2006, October 23). New York, Los Angeles, Olathe? Rent's bite is big in Kansas, too. New York Times, pp. A1, A16.

Standards for defining Metropolitan and Micropolitan Statistical Areas; Notice, 65 Fed. Reg. 82228 (Dec. 27, 2000).

Stewart B. McKinney Homeless Assistance Act, 42 U.S.C. § 11301 et seq. (1987, 1999).

Strong, D. A., Del Grosso, P., Burwick, A., Jethwani, V., Ponza, M., Bhatt, J., Phillips, S., & Scheppke, K. (2005). Rural research needs and data sources for selected human services topics. Princeton, NJ: Mathematica Policy Research.

Substance Abuse and Mental Health Services Administration, Center for Mental Health Services. (2006). Expert panel meeting on rural homelessness. Rockville, MD: Westat.

U.S. Department of Education. (2004). Education for homeless children and youth program: Title VII-B of the McKinney-Vento Homeless Assistance Act, as amended by the No Child Left Behind Act of 2001, non-regulatory guidance. Washington, DC: Author.

U.S. Department of Housing & Urban Development, Office of Policy Development & Research. (1995). Fair market rents for the Section 8 Housing Assistance Payments Program. Washington, DC: Author.

Vermont Department of Social Welfare. (1985). Homelessness in Vermont. Waterbury, VT: Author.

Vermont Department of Social Welfare. (1987). Homelessness in Vermont: A research survey. Waterbury, VT: Author.

Vernez, G., Burnam, M. A., McGlynn, E. A., Trude, S., & Mittman, B. (1988). Review of California's program for the homeless mentally disabled (Rep. No. R-3631-CDMH). Santa Monica, CA: The Rand Corporation.

Wasserman, K. (2006, July). Developing plans to end homelessness in rural communities. Presented at the National Alliance to End Homelessness Annual Conference. Washington, DC.

Wilder Research Center. (2007a). Homeless in Minnesota 2006: At a glance. Saint Paul, MN: Amherst H. Wilder Foundation.

Wilder Research Center. (2007b). Overview of homelessness in Minnesota 2006: Key facts from the statewide survey. Saint Paul, MN: Amherst H. Wilder Foundation.

Wilder Research Center. (1998). Minnesota statewide survey of persons without permanent shelter: Vol II. Unaccompanied youth. St. Paul, MN: Amherst H. Wilder Foundation.