The Spatial Abandonment and Spatial Clustering of

the Elderly in the United States.

INTRODUCTION

Researchers from a wide variety of disciplines agree on the growing importance of the elderly as a demographic group within American society (Graff & Wiseman, 1978; Meyer & Speare 1985). Both the absolute size and the proportion of the population who are aged 65 or over have increased dramatically in the twentieth century. In 1900 approximately 4.1 percent, or 3.1 million people were aged 65 and over; by 1999 the percent elderly was estimated at almost 12.7 or over 34 million people (U.S. Census Bureau, (1999b). As we move into the twenty first century, and the baby boomers reach retirement age, the elderly are likely to represent a higher percentage of our population than ever before. Projections of the percent of the United States population likely to be elderly by the time age structure stabilizes in the first half on the next century, range from 18 to 23 percent, or over 55 million people (McCarthy et al., 1982; Rogers, 1992; Wiseman, 1979). The provision of services for this large and rapidly growing segment of society is one major concern that has spawned a large body of literature pertaining to the elderly (Bohland & Rowles, 1988; Rogers & Woodward, 1988).

Since the 1970s geographers have also begun to focus on a variety of issues relating to older Americans (Bowles, 1980; Golant, 1979; Rowles 1986). Two such issues have involved poverty and migration. After the 'War on Poverty' was declared by Lyndon B. Johnson in 1964 the spatial aspects of impoverishment in the United States came under close scrutiny (Brunn and Wheeler, 1971; Morrill & Wohlenberg; 1971). As the host of social programs aimed at alleviating widespread poverty were implemented, focus shifted to the demographics and characteristics of the poor who were the intended targets of these programs (Hoffman, 1991; Levitan & Shapiro, 1987; Tobin, 1990). What emerged clearly from studies designed to investigate just which groups suffered the most from poverty, was that the big winners in the 'War on Poverty' were the elderly (Duncan et al., 1986; Duncan & Smith, 1989; Gibbs, 1993; Haveman, 1987; Hobbs & Damon, 1996). By the mid 1980s Americans aged 65 and over, once a severely impoverished group, were enjoying relative affluence compared to African-Americans and children (Duncan et al., 1986; Greenstein, 1986; Tobin, 1990). Thus, as we move toward the new millennium most research pertaining to poverty does not focus on the elderly.

The same lack of focus on older Americans is not apparent in studies relating to migration. People 65 and over are seen as increasingly mobile (Bohland & Rowles, 1988). Streams of elderly migrants, with destinations in the Sunbelt, have been documented by many researchers, as have a variety of more localized movements (Bryant & El-Attar, 1984; Cebula, 1985; Flynn, 1980; Golant, 1987; Longino, 1982). It seems that in the later decades of this century, retirement has become increasingly likely to precipitate a change in residence (Chevan & Fischer, 1979; Flynn et al., 1985).

Within the many discussions pertaining to elderly migration there has been frequent reference to the role of income and affluence. Researchers appear to agree that among older Americans, affluence is likely to be positively related to migration, while lack of resources constitutes a barrier to such movement (Barsby and Cox, 1975; Graff & Wiseman, 1978; Khraif, 1989; Rowles, 1986; Warnes, 1992; Zelinsky, 1971). What is missing from previous study is an attempt to identify possible spatial consequences of the link between affluence and migration, and residential inertia and lack of financial resources. The purpose of this paper is to explore the geographies of elderly affluence and poverty, and their possible relationship to migration versus the aging-in-place of older Americans. Simply put, are there regions where the elderly poor have become an abandoned group who lack the resources to move as others within the population move out, in contrast to concentrations of affluent older Americans that result from migrations made possible by access to financial resources.

BACKGROUND

Poverty and the Elderly

Despite prosperous economic times in the United States poverty remains a persistent and widespread problem. In 1997 13.3 percent of the U.S. population, or some 35.6 million people lived below the official poverty line. Although 13.3 percent poverty represents the lowest rate since 1989, the total number of poor Americans still remains greater than at any time during the 1970s or 1980s. Clearly the problem of a large economically disadvantaged segment of the population has not been solved as we approach the dawn of the 21^st century. (Dalaker & Naifeh, 1997).

When the composition of the impoverished population is scrutinized however, it is clear that the elderly as a group have made large gains in terms of poverty status since the implementation of the Social Security Act of 1935 and Johnson's 'War on Poverty' declared in 1964 (Anderson, 1990; Gibbs, 1993; Hobbs & Damon, 1996; Paul, 1988). Social welfare programs have treated the elderly as part of the 'worthy poor' and increased transfer payments have been the result. In addition, improvements in private income, such as that derived from pensions, have also meant good financial news for America's senior citizens. As a result of both social welfare and improved private income the elderly are the only major group to have enjoyed a decrease in the poverty rate. The proportion of the elderly who are considered poor has certainly dropped dramatically. In 1959 over 35 percent of all elderly persons were poor; by 1997 that percentage was only 10.5. In fact, despite the rapidly growing number of persons aged 65 and over, the actual number of impoverished elderly has remained largely stable (Axinn & Stern, 1988).

Thus, the elderly, as a group, are touted as the social welfare success story of this century. A contrast to other groups such as African-Americans, children, and people within female headed households who continue to suffer high rates of poverty (Clark et al., 1988; Duncan et al, 1986). As the reduction in the elderly poverty rate is hailed as a major victory within the war on poverty, it is tempting to imagine all senior citizens enjoying their golden years with at least a comfortable income. Such comfort, however, is still not available to a significant group of older Americans (Anderson, 1990).

The 10.5 percent of the elderly who remain poor represent over 3.3 million people (Dalaker & Naifeh, 1997). In addition, the elderly population is far from homogeneous and certain subgroups are at high risk of impoverishment. The 'oldest old' are one group that suffer higher rates of poverty than the rates for the 65 and over group would suggest (Paul, 1988). In 1992, the poverty rate among people 65 to 74 years of age was 10.7 percent, in sharp contrast to the 15.3 percent rate among 75 to 84 year olds, and the 19.8 percent rate among people aged 85 or older. It seems clear that the older senior citizens become the greater their risk of falling into poverty. In fact, rates for the elderly aged over 84 are not significantly different from those for children, who many researchers point to as the most seriously disadvantaged group (Hobbs & Damon, 1996). This fact is particularly disturbing when one considers that an increasing percentage of the elderly are the 'oldest old' (Axinn & Stern, 1988).

Also at high risk of being poor are elderly women, elderly African-Americans, and seniors with significant health problems and disabilities. In 1992, the poverty rate of elderly men was 8.9 percent; the rate for women was 15.7 percent (Hobbs & Damon, 1996). While women make up approximately 58 percent of the elderly population, they comprised over 71 percent of the elderly poor in 1992. Often female poverty among the elderly is linked to widowhood and to advanced age (Anonymous, 1992; Clark et al., 1988). As Clark et al. remark 'the story of the old and poor is mainly about women' (Clark et al., 1988 p.8 ).

Much as in the general population race is related to risk of impoverishment. In 1992 10.9 percent of the white elderly were poor, while rates for both Hispanic and black elderly were much higher at 22 and 33.3 percent respectively. In addition to the links between gender, race and poverty detailed above, health status is also related to impoverishment. The chronically ill and disabled, who must incur ongoing health related costs and who cannot take advantage of economic opportunity, often fall into poverty (Duncan & Smith, 1989).

Thus, a picture emerges of elderly poverty that is in many ways a microcosm of the larger society. There seems to be a polarization within elderly Americans with affluent white males or married couples who enjoy good health and are still relatively young on one end of the financial spectrum, and the oldest old, single women, and racial minorities on the other. Two additional points are also worthy of note. The official poverty line is drawn differently for persons aged 65 and over than it is for other people. The elderly are deemed to need less income to stay out of poverty since their nutritional needs are lower and they have accumulated assets over a lifetime (Clark et al., 1988). In 1998 the poverty line for a single person aged under 65 was $8,480 and for a two person household it was $10,915; the equivalent poverty lines for persons aged 65 and over were $7,818 and $9,853 (U.S. Census Bureau, 1999a). As Clark et al., 1988 note if the same poverty line were used for older Americans as for other people, the elderly poverty rate would have been 15.2 rather than 12.6 percent in 1992. Clearly the elderly are more likely to be close to the poverty line than other people. Of the 12.3 million 'near poor', with 1992 incomes between 100 - 125 percent of the poverty level, 18.9 percent were elderly. The second point that is worthy of attention is the fact that the elderly poor are more likely than other impoverished groups to be chronically poor. Once in poverty their chances of escape are not good (Coe, 1988; Eller, 1996; Hobbs & Damon, 1996).

Migration and the Geography of the Elderly

The basics of elderly migration have been well documented and contribute to the overall geography of the elderly in the United States. The elderly population is not only growing but is becoming increasingly mobile (Bohland & Rowles, 1988; Flynn et al., 1985; Warnes, 1992). While it is true that the elderly migrate less frequently than younger groups the impacts of their movement can be substantial (Biggar, 1980b; Carter, 1994; Graff & Wiseman, 1978). Looking at the state scale, the numbers of elderly leaving states largely reflects general population levels, with New York leading the way in terms of sheer number of out-migrants. The primary destinations for these interstate elderly migrants, in contrast to the dispersed spatial pattern of out-migration, are highly spatially clustered (Biggar, 1979; Flynn, 1980; Graff & Wiseman, 1990). Wiseman, 1986 recognized two national streams into Florida and California, five regional streams into Missouri-Arkansas, Washington-Oregon, New York, New Jersey, and Ohio., and two emerging streams to Arizona and Texas. Others have more generally characterized these flows as a movement toward the Sunbelt and the West (Biggar, 1980a), and have pointed to the emergence of new retirement areas in North Carolina, Georgia, South Carolina, and New Mexico (Bohland & Rowles, 1988). Clearly these interstate movements of the elderly, all other things being equal, means a concentration of older Americans within destination areas and a reduced concentration in origin areas.

Added to these interstate migrations of older Americans, it is recognized that many moves by the elderly are more local in nature. Senior citizens, it is said, have led the way in the migration from metropolitan counties to non-metropolitan counties that began in the 1970s (Golant, 1987; Longino, 1980, 1982; Meyer, 1987). In addition they have been an integral part of population flows from central cities toward the suburbs (Bryant & El-Attar, 1984; Fuguitt & Tordella, 1980; Golant, 1987). Thus, seniors have contributed to their own spatial distribution by moving to the Sunbelt, by moving to non-metropolitan counties, and by leaving central cities for more suburban locations. The consequence of all these types of moves is that the change in elderly population for counties has been very variable, ranging from a loss of 16.5 percent to a gain of 13.5 percent (Bohland & Rowles, 1988).

Beyond the migration of the elderly, the concentration of post-retirement aged people is impacted by the migration of other people, as well as historic population patterns that are retained as people age-in-place (Bohland and Rowles, 1988; Graff & Wiseman, 1978; Wiseman, 1986). In many places out-migration and in-migration on non-elderly persons may be more important in explaining elderly concentration than any movement of the elderly themselves (Bohland & Rowles, 1988; Fuguitt & Tordella, 1980). Certainly heavy in-migration of the relatively young into the same destinations as the elderly is likely to result in no apparent concentration of elderly population. Similarly, out-migration of the non-elderly from places where the aging population is stable may result in a concentration of the elderly population. In essence the young have been leaving the elderly behind in some rural areas, as they have been attracted by the expanding economies often to be found in the Sunbelt and in suburban counties that are part of large metropolitan areas (Graff & Wiseman, 1978; 1990; Wiseman, 1979; 1986).

Clearly, as Graff & Wiseman (1978) indicate the geography and spatial concentration of America's elderly population is the result of several processes; out-migration of the old, in-migration of the old, out-migration of the young, in-migration of the young, as well as both the aging-in-place and dying-in-place of the elderly. These processes act in concert, but may all be linked by economic factors (Golant, 1979).

Links Between Well Being and Elderly Mobility

There appears to be a consensus amongst researchers concerning both the primary motivation for elderly migration, and the link between well-being and the propensity to move. Unlike the young, who often migrate toward expanding economies in search of jobs and improved income, the promise of higher income is not what motivates the elderly to migrate (Cebula, 1974; 1985; Chevan and Fischer, 1979). Given the often fixed incomes of the elderly, consumption is the key to understanding elderly migration (Biggar, 1980a; McCarthy et al., 1982). Lower living costs may attract the elderly since it means that their income can buy more (Fournier et al, 1988; Wiseman, 1979). In addition, an expanded availability of services and recreational opportunities are often a major pull factor for the elderly (Fuguitt & Tordella, 1980; Graff & Wiseman, 1978). It is not enough that a low cost of living make fixed incomes go farther; many seniors want to spend their money on amenities and recreation that just may not be available in their places of origin (Khraif, 1989).

The ability to seek out consumption opportunities and lower living costs demands a certain level of affluence (Warnes, 1992; Zelinsky, 1971). Those with limited financial resources cannot afford to move, and even if they could, the availability of amenities and recreational opportunities become frustrations if one lacks the financial resources to consume them (Lee, 1980). Thus, researchers agree that elderly mobility is directly linked to higher income and affluence (Barsby & Cox, 1975; Golant, 1972). The social characteristics of elderly movers are associated with income, with the younger elderly, married couples, and those in good health most likely to migrate (Biggar, 1980b; Chevan & Fischer, 1979; Meyer, 1987; Rogers, 1992). There are also less direct connections, between retirement migrations and a life-time pattern of relative economic comfort. People who have commanded a good income during their working years are more likely to have made repeated migrations over their lifetime, as they went to college, married, and followed job opportunities. To these people a retirement migration is just one in a series of moves related to life's developmental stages (Chevan & Fischer, 1979; Litwak & Longino, 1987; Meyer and Speare, 1985; Rogers, 1992). In addition, people who have had a good income in the past have accumulated knowledge of the various retirement options as they traveled for a variety of reasons such as business or vacations (Graff & Wiseman, 1978; Lee, 1980).

The extensive literature available gives us a clear picture of the geography of the elderly that is a result of the various processes mentioned. It is no surprise to see a map on which counties in Florida stand out as having a high percentage of old people due to in-migration of the elderly, or in the mid-west due to elderly aging-in-place plus out-migration of the young (Graff and Wiseman, 1978; Wiseman, 1986). What is less clear is whether, within this overall geography of elderly concentration, there are sub-regions of the poor, spatially abandoned elderly, versus affluent regions were the elderly with sufficient resources have chosen to congregate. This issue has not been explored despite the acceptance that the elderly who migrate and those that do not tend to have different socioeconomic characteristics (Rogers & Woodward, 1988).

METHODS

Data relating to age, elderly poverty, population change, and socioeconomics available from the U.S. Census Bureau at the county level, were utilized in this study. ArcView, a geographic information system (GIS) was used to construct a series of maps to display the spatial distribution of elderly concentrations, the spatial distribution in terms of absolute numbers of the elderly, the percentage of the elderly who are poor, and counties that have lost or gained significant population from 1980 to 1990. The analytical power of the GIS is used to find possible spatial relationships between these age, elderly poverty, and population change characteristics. A Map is created that displays: 1) counties with high concentrations of elderly population, that suffer high average poverty rates, and that have lost significant population; 2) counties with high concentrations of elderly population, that enjoy much lower than average poverty rates, and that have gained significant population. The spatial distribution of these two groups of counties may indicate the existence of a geography of abandonment where the poor elderly are concentrated, versus a geographic clustering of the more affluent elderly. Further analysis, using socioeconomic data and applying Analysis of Variance techniques, was conducted in an attempt to determine if the two county groupings have any distinct profile. Also explored was the possibility that distinct subgroups of counties may exist within the two basic county groupings; this was accomplished by conducting a Principle Components Analysis followed by a cluster analysis. All maps produced used means (M) and standard deviations (SD) that characterized the data, to arrive at mapping intervals. Table 1 provides information on these limits used when constructing the maps.

Table 1. Limits Used to Construct the Maps.

RESULTS

The initial series of maps produced really reveal few surprises. Mapping the percentage of the population of elderly population in 1990 shows a distinct pattern in terms concentration of the elderly. At the spatial scale of the county, the average percent of population aged 65 or older was 14.95, and the standard deviation from this mean was 4.34 (Table 1). Counties that deviate by more than one standard deviation above the mean are clustered in a broad band in the Central United States from Texas to North Dakota, and in several coherent areas including Florida, the central Ozarks, central Montana, Arizona, and coastal Pacific Northwest. Most of the eastern half of the United States lacks any clear concentration of elderly, as do large portions of the central west and extreme southwest (Figure 1). This spatial pattern of elderly concentration comes as no surprise given the results of previous research discussed earlier. What is somewhat significant is that when viewing a map which locates the absolute numbers of elderly Americans (using categories of <M, +1 SD, +2 SD, and greater than +2SD) the elderly population does not closely mirror the spatial distribution of the population as a whole (Figure 2). While the northeast does stand out somewhat, what also shows clearly are the large numbers of elderly Americas who live in the Seattle, San Francisco, southern California/Utah, and Florida areas as well as in coastal sections of New England. In fact, the 83 counties (2.6 percent of 3140 U.S. counties) with elderly populations greater than 70,733 (over 2 SD above the mean), are home to more than 38 percent of the population aged 65 and over. America's elderly population is concentrated not only in terms of percent but also in absolute terms.

Figure 1. Percent of U.S. County population Aged 65 and Over, 1990

Figure 2. Number of People by U.S. County Aged 65 and Over, 1990

With respect to elderly poverty, the spatial distribution observed is consistent the geography of the poor within the general population. General poverty regions identified by Brunn and Wheeler (1971) and later by Shaw (1996) are also regions with the highest rates of elderly poverty (Figure 3). Cores of elderly poverty are clearly spatially delineated, and are centered in northeast Arizona/northwest New Mexico, the Texas border region, the Dakotas, Appalachian Kentucky, and the southern coastal plain in a broad sweep from Louisiana to North Carolina (Figure 3). Relatively subtle differences between the concentration of the impoverished elderly versus the poor general population may be worthy of note. In particular, the poverty region that spans the coastal plain draws in some additional areas when only the poor elderly are considered. Many counties in eastern Arkansas, the Ozarks of Missouri/Arkansas, western Louisiana, and southeastern Oklahoma suffer high rates of elderly poverty. These are areas normally considered to be on the fringes of rural poverty regions, but with the spatial contraction of poverty in recent decades, no longer a part of general poverty cores.

Figure 3. Percent of U.S. County Population Aged 65 and Over who are Poor, 1990

Substantial population loss between 1980 and 1990 was primarily an experience confined to the central United States (Figure 4). Counties that lost more than 12.53 percent of their population (over 1 SD from the mean) can be seen in a broad band from Montana/North Dakota south to the Texas panhandle. Interestingly some additional small areas of loss can be seen at the heart of the poverty cores in New Mexico and in particular along the lower Mississippi valley. These areas of loss are largely rural and offer limited economic opportunity.

Figure 4. Percent Population Loss for U.S. Counties, 1980-1990

Population gain from 1980 to 1990 also shows a clear spatial pattern (Figure 5). Gains in population of over 20.77 percent (over 1 SD from the mean) are concentrated in the southwest, in Florida, and in some metropolitan areas. Among metropolitan and surrounding areas that stand out as gaining significant population are Seattle, Salt Lake City, Denver, Albuquerque, Minneapolis-St. Paul, St. Louis, Nashville, Atlanta, and a cluster of metropolitan counties in the Washington D.C. area. In contrast to regions that lost substantial population, regions that made significant gains are often in the sunbelt, and are associated with the growth economies of metropolitan areas.

Figure 5. Percent Population Gain for U.S. Counties, 1980-1990

Viewing this first series of maps does confirm conventional wisdom in terms of the spatial distribution of the elderly, the elderly poor, and population change. However, they do nothing to aid in identification of regions where elderly concentration, elderly poverty, and population loss coincide or where there is a concentration of elderly population with low poverty rates and population gain. Figure 6 displays all counties in the United States that meet three criteria. Counties with both percent of the population elderly and percent old in poverty greater then 1 SD from the mean (25.36 and 19.29 respectively) that also lost population between 1980 and 1990 are identified as a group. Nineteen counties in the conterminous United States meet the three criteria established. These counties may be said to represent a geography of spatial abandonment, where the poor elderly have been forced to age-in-place while other groups have left, presumably to seek economic opportunity elsewhere. Two geographic clusters stand out, and may be seen bounded by lines on Figure 6. The largest region of elderly abandonment can be seen spanning much of eastern Texas, crossing into some southern-most counties in Oklahoma. Of the 23 counties identified, 15 are included in this cluster. The second cluster consists of six counties in northern Arkansas, southern Missouri, and the western-most tip of Kentucky. Only two counties that meet the criteria to be designated as areas of elderly abandonment exist outside of the two clusters identified; they are single counties in North Carolina, and North Dakota. Table 2 shows the number of counties, by state, that are a part of this poor group.

Table 2. Counties by State Included in Poor Versus Non-Poor Groupings.

Figure 6 also displays counties where there is a concentration of elderly population with low poverty rates and population gain. Counties that again have the percent of elderly population greater then 1 SD from the mean (25.36), but that also experience poverty rates less that 1 SD below the mean (8.84) and have gained at least 20.77 percent in total population between 1980 and 1990 are grouped. The 40 counties that meet these criteria may be said to represent clustering of the non-poor elderly. One large and coherent cluster of the non-poor elderly stands out, and is located in peninsular Florida. Of the 40 counties identified 18 are included in this cluster. The remaining 22 counties are scattered across the central United States north of Texas, with additional small nodes of counties along the northwest coast and in the northeast (Table 2). It is interesting to note that only North Dakota has counties within both the non-poor elderly with population gain, and poor elderly with population loss groupings.

Figure 6. High Elderly-High Elderly Poverty-Population Loss and High Elderly-Low Elderly Poverty-Population Gain U.S. County Groupings

In order to investigate if basic differences exist between these two groups of counties - the 23 counties that represent the spatial abandonment of the elderly poor versus the 40 that represent the spatial clustering of the elderly non-poor - a variety of socioeconomic variables were explored. Some distinctions between the two groups of counties are relatively easy to observe. In particular the counties where the elderly poor are clustered are overwhelmingly rural, in contrast to the significant number of counties where the non-poor are clustered that are part of metropolitan areas. On the 23 counties in the poor group, none are metropolitan, and only five are even adjacent to a metropolitan area. Instead these counties are relatively isolated and rural. Of the 40 counties in the non-poor group 18 are part of metropolitan areas, and another 8 are adjacent to such areas. Other distinctions are less easy to see via simple observation, and so Analysis of Variance (ANOVA) testing was applied to data relating to age, education, race, gender, economic well-being, health, and type of place.

The ANOVA testing revealed that significant differences between the two groups of counties exist at the .05 level in terms of population density, percent white, percent black, percent widowed, the death rate, physician rate, percent of the population receiving supplemental social security benefits, high school graduation rate, college graduation rate, median family income, percent of very low income families, percent of high income families, median house value, per capita retail sales, per capita income, and percent of the population that may be thought of as the oldest of the old. No significant difference was found in the male-female ratio, percent of the population in nursing homes, social security participation rate, unemployment, or percent of the population who live on farms. Thus, counties that constitute a geography of the spatial abandonment of the elderly poor are characteristically rural, have low population densities, have a relatively low percentage of white and a high percentage of black residents when compared to counties in the non-poor group, have a relatively high percentage of widows, suffer high death rates, have limited access to health care, are home to many people who receive supplemental social security benefits, have a poorly educated population, have generally low family and personal income, have low house values, have a less than vibrant retail sector, and are home to disproportionate numbers of the oldest old. Counties identified as belonging to the non-poor group display an opposite tendency in terms of the characteristics listed above.

Despite the basic distinctions between the two groups of counties, an examination of the socioeconomic characteristics of counties that belong within the groups leads to the suspicion that some meaningful sub-groupings may exist. For example, it seems to be mainly the Florida counties that account for the metropolitan nature of the non-poor county group. In order to explore possible sub-groupings within the poor and non-poor county groups a Principle Components Analysis (PCA) was applied to the same socioeconomic data as was used in the ANOVA, and a subsequent cluster analysis was performed on identified factors. It was necessary to apply PCA to the data prior to any cluster analysis, in order to extract unrelated dimensions within the data set, since the 25 variables used are likely to suffer from problems associated with multicollinearity.

Applying a PCA to the data resulted in 5 factors with eigenvalues over 1. These five factors contained over 80 percent of the information embraced by the 25 original socioeconomic variables. Based on the loadings the 5 factors were interpreted to represent income and education, extreme old age and infirmity, gender and dependency, race, and polarity of income. It is not the interpretation of just what these factors represent that is of primary interest, but more if counties cluster in terms of these latent dimensions within the socioeconomic data. Taking the 7 cluster level of an hierarchical cluster analysis reveals some interesting associations of counties.

The 23 counties that comprise the poor group are basically divided into two subgroups by the cluster analysis (Figure 7c and 7g). Seven counties in east Texas are grouped together as similar socioeconomically, but this cluster also includes two non-poor counties, one in Florida (Collier) and one in Washington (San Juan). The remaining poor counties form a cluster seen as similar socioeconomically that embrace the west Texas/Oklahoma and Arkansas/Missouri counties.

Within the non-poor county group there is more variation than within the poor group, and several sub-groupings of counties are apparent. Two one counties clusters indicate that both Glades and Pinellas counties in Florida are unique (Figure 7a and 7b). While both are part of the non-poor county group they do not share socioeconomic characteristics with other group members or with each other. Another sub-grouping of counties includes 4 south Florida counties as well as the 2 New Jersey counties and 1 Massachusetts county focused upon in this study (Figure 7d). The remaining 11 Florida counties form a group with similar characteristics; also included in this cluster are the northwestern counties in Oregon and Washington as well the single counties in Colorado, Pennsylvania, and Wisconsin that are part of the non-poor group (Figure 7f). A final cluster of counties is identified by the analysis that are located in the central United states from northern New Mexico to North Dakota. Interestingly the two North Dakotan counties are both placed in this cluster despite one being designated poor and one non-poor.

Figure 7. Seven Clusters of Counties

DISCUSSION

The research presented here indicates that regions of both poor, elderly abandonment and where the non-poor cluster can be envisaged. While the number of counties that display concentrations of elderly population, high rates of elderly poverty, and population loss are limited, they display a distinct geography. The eastern two-thirds of Texas is the primary area of abandonment with a secondary region along the Arkansas/Missouri border area. It is perhaps surprising that areas in the central Unites States north of Texas, that are the major region of population loss as well as an aging population, are not a part of the region where the elderly poor can be said to be spatially abandoned. The 23 counties that are included in this poor group share a socioeconomic profile that is fundamentally different from that for the non-poor county grouping. Here the elderly tend to be rural, perhaps minority, are poorly educated, suffer from ill health, tend to be among the oldest old, and of course lack income and assets and are forced to reply on transfer payments. However, sub-clusters within this group of counties reveal that the Texan counties located toward the west share more socioeconomically with the Arkansas/Missouri counties than with eastern-most Texan counties. While it should be noted the actual numbers of poor elderly in these 23 counties is not large and represents a small proportion of America's elderly poor, they do constitute a meaningful group within the counties involved. Within these counties 28.8 of residents aged over 65 are poor, and these people make up over 6 percent of the total county populations. Combined with the recent history of population loss this must be cause concern at the local, if not the national level.

Clearly the most significant number of counties with a concentration of elderly population, low rates of elderly poverty, and population gain are to be found in peninsular Florida. However, areas in the central United States, north of Texas, as well as coastal areas in the northwest are a part of this grouping. The concentration of the non-poor elderly in Florida and in the northwest may be explained by a search for amenities. These counties are often urban and offer water-based recreation. Why a smattering of counties in the upper central portion of the U. S. are part of the non-poor grouping is more difficult to explain since these are mostly rural counties. Perhaps the concentration of non-poor elderly to these counties represent more local migrations, that allow elders to stay close to family members. Exploration of the reasons non-poor elderly population reside in these counties might be an interesting topic for further research. They are certainly revealed to be distinct from other counties located in regions traditionally associated with spatial concentration of the affluent elderly.

REFERENCES CITED

Anderson, G. M. (1990). Old and Poor in the U.S.A. America, 163 (13), 328-329+, November 3.

Anonymous. (1992). Millions of Older Women Live in Poverty. News for You, 40 (38), 1, September 23.

Axinn, J. & Stern, M. J. (1988). Dependency and Poverty. Old Problems in a New World. Lexington, Massachusetts: Lexington Books.

Barsby, S. L. & Cox, R. C. (1975). Interstate Migration of the Elderly. Lexington, Massachusetts: Lexington Books.

Biggar, J.C. (1979). The Sunning of America: Migration to the Sunbelt. Population Bulletin, 34 (1), 3-42.

Biggar, J. C. (1980a). Reassessing Elderly Sunbelt Migration. Research on Aging, 2 (2), 177-180.

Biggar, J. C. (1980b) Who Moved Among the Elderly. Research on Aging, 2 (1), 73-91.

Bohland, J. R. & Rowles, G. D. (1988). The Significance of Elderly Migration to Changes in Elderly Population Concentration in the United States: 1960-1980. Journal of Gerontology: Social Sciences, 43 (5), S145-152.

Bowles, G. K. (1980). Age Migration in the United States. Research on Aging, 2 (2), 137-140.

Brunn, S.D. & Wheeler, J.O. (1971). Spatial Dimensions of Poverty in the United States. Geografiska Annaler, 53B (1), 6-15.

Bryant, E. S. & El-Attar, M. (1984). Migration and redistribution of the Elderly: A Challenge to Community Services. The Gerontologist, 24 (6), 634-640.

Carter, J. (1994). Elderly Cohort Migration Patterns. New York: Garland Publishing, Inc.

Cebula, R. J. (1974). The Quality of Life and Migration of the Elderly. Review of Regional Studies, 4 (1), 62-68, Spring.

Cebula, R. J. (1985). Living Costs, the Quality of Life, and the "Sunbelt" vs "Frostbelt" Battle in the United States. Review of Regional Studies, 15 (3), 49-53, Fall.

Chevan, A. & Fischer, L. R. (1979). Retirement and Interstate Migration. Social Forces, 57 (4), 1365-1380, June.

Clark, W. F., Pelham, A. O., & Clark, M. L. (1988). Old and Poor. Lexington, Massachusetts: Lexington Books.

Coe, R. D. (1988). A Longitudinal Examination of Poverty in the Elderly Years. The Gerontologist, 28 (4), 540-544, August.

Dalaker, J., & Neifeh, M. (1997). Poverty in the United States. Washington D.C.: Bureau of the Census, Current Population Report P60-201.

Duncan, G. J., Hill, M., & Rodgers, W. (1986). The Changing Fortunes of Young and Old. American Demographics, 8 (8), 26-33, August.

Duncan, G. J., & Smith, K. R. (1989). The Rising Affluence of the Elderly: How Far, How Fair, and How Frail? Annual Review of Sociology, 15, 261-289.

Eller, T.J. (1996). Who Stays Poor and Who Doesn't. Dynamics of Economic Well-Being:

Poverty 1992-1993. Washington D.C.: Bureau of the Census, Current Population Report P70-55.

Flynn, C. B. (1980). General Versus Aged Interstate Migration, 1965-1970. Research on Aging, 2 (2), 165-176, June.

Flynn, C. B., Longino, C. F., Wiseman, R. F., & Biggar, J. C. (1985). The Redistribution of America's Older Population: Major National Migtration Patterns for Three Census Decades. The Gerontologist, 25 (3), 292-296.

Fournier, G. M., Rasmussen, D. W., & Serow, W. J. (1988). Elderly Migration as a Response to Economic Incentives. Social Science Quarterly, 69 (2), 245-260.

Fuguitt, G. V. & Tordella, S. J. (1980). Elderly Net Migration. The New Trend of Non-Metropolitan Population Change. Research on Aging, 2 (2), 191-204.

Gibbs, W. W. (1993). Chronologically Privileged. Scientific American, 269 (4), 107, October.

Golant, S. M. (1972). The Residential Location and Spatial Behavior of the Elderly. A Canadian Example. Chicago: University of Chicago, Department of Geography Research Paper No. 143.

Golant, S. M. (1979). Rationale for Geographic Perspecives on Aging. In Location and Environment of Elderly Population. Ed. S. M. Golant. New York: Halsted Press, John Wiley and Sons, 1-16.

Golant, S. M. (1987). Residential Moves by Elderly Persons to U.S. Central Cities, Suburbs, and Rural Areas. Journal of Gerontology, 42 (5), 534-539.

Graff, T. O. & Wiseman, R. F. (1978). Changing Concentrations of Older Americans. The Geographical Review, 68 (4), 379-393, October.

Graff, T. O. & Wiseman, R.F. (1990). Changing Pattern of Retirement Counties Since 1965. The Geographical Review, 80, 239-251.

Greenstein, R. (1986). Numbers and Need: The Outlook for Working Families and the Poor. Chicago: Midwest Regional Conference, Agenda '87: Revolutionizing State Economic Policy, November 21-22.

Haveman, R. H. (1987). Poverty Policy and Poverty Research. Madison. The Great Society and the Social Sciences. Wisconsin: University of Wisconsin Press.

Hobbs, F. B., & Damon, B. L. (1996). 65+ in the United States. Washington D.C: Bureau of the Census, Current Population Report, P23-190, April.

Hoffman, E. P. (1991). Aid to Families with Dependent Children and Female Poverty. Growth and Change, 22 (2), 36-47.

Khraif, R. (1989). The Interstate Migration and Destination Choice of the Elderly in the United States. Indiana University: Department of Geography, Ph. D. Dissertation, February.

Lee, E. S. (1980). Migration of the Aged. Research on Aging, 2 (2), 131-135, June.

Levitan, S. A. & Shapiro, I. (1987). Working But Poor. Baltimore, Maryland: Johns Hopkins University Press.

Litwak, E. & Longino, C. F. (1987). Migration Patterns Among the Elderly: A Developmental Perspective. The Gerontologist, 27 (3), 266-272.

Longino, C. F. (1980). Residential Location of Older People. Research on Aging, 2 (2), 205-216, June.

Longino, C. F. (1982). Changing Aged Nonmetropolitan Migration Patterns, 1955 to 1960 and 1965 to 1970. Journal of Gerontology, 37 (2), 228-234.

McCarthy, K. F., Abrahamse, A., & Hubay, C. (1982). The Changing Geographic Distribution of the Elderly. Santa Monica, California: Rand R-2895 NIA, March.

Meyer, J. W. (1987). County Characteristics and Elderly Net Migration Rates. A Three-Decade Regional Analysis. Research on Aging, 9 (3), 441-452.

Meyer, J. W. & Speare, A. (1985). Distinctively Elderly Mobility: Types and Determinants. Economic Geography, 61, 79-88.

Morrill, R.L. & Wohlenberg, E. H. (1971). The Geography of Poverty in the United States. New York: McGraw-Hill Book Co., Problems Series.

Paul, A. (1988). High Rates of Poverty Afflict 'Oldest Old'. Chronicle of Higher Education, 34 (37), A15, May 25.

Rogers, A. (1992). Elderly Migration and Population Redistribution in the United States. In Elderly Migration and Population Redistribution. A Comparative Study. Ed. A. Rogers. London: Belhaven Press, 226-248.

Rogers, A. & Woodward, J. (1988). The Sources of Regional Elderly Population Growth: Migration and Aging-in-Place. The Professional Geographer, 40 (4), 450-459.

Rowles, G. D. (1986). The Geography of Aging and the Aged: Toward an Integrated Perspective. Progress in Human Geography, 10 (4), 511-539, December.

Shaw, W. (1996). The Geography of United States Poverty. Patterns of Deprivation, 1980-1990. New York: Garland Publishing, Inc.

Tobin, J. (1990). The Poverty Problem: 1964-1989. Focus (University of Wisconsin-Madison, Institute for Research on Poverty), 12 (3), 6-7, Spring.

Warnes, A. M. (1992). Age-Related Variation and Temporal Change in Elderly Migration. In Elderly Migration and Population Redistribution. A Comparative Study. Ed. A. Rogers. London: Belhaven Press, 35-55.

Wiseman, R. F. (1979). Regional Patterns of Elderly Concentration and Migration. In Location and Environment of Elderly Population. Ed. S. M. Golant. New York: Halsted Press, John Wiley and Sons, 21-36.

Wiseman, R. F. (1986). Migration of Older Americans. In Housing an Aging Society. Issues, Alternatives, and Policy. Eds R. J. Newcomer, M. P. Lawton, T. O. Byerts. New York: Van Nostrand Reinhold Company, 69-82.

U.S. Census Bureau. (1999a). Poverty Thresholds: 1998. Internet WWW page at, URL: http://www.census.gov/hhes/poverty/threshld/thresh98.html, May 25.

U.S. Census Bureau. (1999b). Poverty Thresholds: 1998. Internet WWW page at, URL:http://www.census.gov/population/estimates/nation/intfile2-1.txt, June 4.

Zelinsky, W. (1971). The Hypothesis of the Mobility Transition. The Geographical Review, 61 (2), 219-249, April.