Slicing and Dicing Counties in Research on Rural Aging
Download
Report
Transcript Slicing and Dicing Counties in Research on Rural Aging
Rural Aging Research and Change in
Rural Counties: An Emerging Research
Agenda
Jim Mitchell
Professor of Sociology and Family Medicine
Director, Center on Aging, Brody School of Medicine
Associate Director, UNC Institute on Aging
Presentation Goals
• To review the use of Rural (residence) as an
independent variable in aging research published since
2000 to illustrate the importance of definition and
measurement.
• To assess differential change in selected demographic
and economic characteristics of rural counties in 4
eastern states over a 40-year period. Rural counties are
more different than represented by federal definitions?
• To explore the notion of rural according to definitions
used by federal agencies and its implication for
secondary data analysis.
• To suggest a strategy to promote research on aging
farmers.
First Census Definition:
All territory, population,
and housing units located
outside of Urban
Clusters…core census
block groups with
population densities of at
least 1,000 people per
square mile and
surrounding census blocks
with at least 500 people
per square mile.
Second Census Definition:
Territory, population, and housing units located
outside of Urbanized Areas… continuously builtup with populations of 50,000 or more people or
places and areas with population densities
greater than 1,000 inhabitants per square mile
(population size within designated area and
density).
Third Census Definition:
Territory, population, and housing units located outside of
Urban Places… either incorporated places or censusdesignated places with 2,500 or more inhabitants.
Office of Management and Budget (OMB)
Definition (pertinent to economic matters and
resource allocation):
Outside of a Metropolitan Area …a metropolitan
influence region with a city with 50,000 or more
inhabitants or a total metropolitan population of
100,000 or more (75,000 or more in New
England) people. Adjacent counties are included
if either fifty percent of their population falls in
the urbanized area surrounding the core city or a
substantial proportion of their workers commute
for employment to the central urban location.
Limitations of Census and OMB
Definitions…
• Metropolitan counties (OMB) have large rural
areas within their bounds, particularly in Western
regions of the country.
• Many Census rural areas include large numbers of
residents living close to places with population
densities as high as 999 people per square mile.
Administration on Aging contribution…
Urban Zip Areas…those with one percent
or more (of population) within an
urbanized area (Census, we presume) and
affiliation with a place of 20,000 to 49,000
residents outside an urbanized area.
USDA Beale Codes…Metropolitan, Urban, and
Rural Counties with Adjacency
• Metropolitan…counties with 250,000+ population.
• Urban…Counties with 20,000+ population (adjacent or
non-adjacent top metro counties).
• Urban…Counties with 2,500+ population (adjacent or
non-adjacent to metro)
• Rural…Counties with fewer than 2,500 population
(adjacent or non-adjacent to metro)
Scheme revised recently to include micropolitan counties
with urban concentrations of 10,000+(up to 20,000?)
people.
USDA Economic Research Service Non-Metro
County Classification (2004)
• Economic Dependence (farming, mining, manufacturing,
government, services, and non-specialized). Farmingdependent counties are those in which 15 percent or
more of earnings or employment are attributable to
farming.
• Policy (retirement destination, federal lands, commuting,
persistent poverty, and transfers dependent) activity.
Useful if continuous, non-categorical measure (e.g.,
continuum of farming dependent)
Rural as an independent variable in
research on aging…
Reviewed 36 full-text articles published from 2000 to
2006 available electronically via the GSA allelectronic journal data bank (including GSA and
other journals). All articles featured the term RURAL
in their titles or their abstracts.
Articles According to Rural
Definition/Measurement
• 6 articles provided clear definition and
measurement (Census and OMB definitions
used 5 of 6 times).
• 5 articles included descriptions of secondary
data used and definition/measurement of rural
could be derived.
• 25 articles featured neither definition nor
measurement description of rural.
In sum, 30 of 36 articles featured neither definition nor
measurement description of rural residence.
Rural County Variability…
Premise or “null hypothesis”--- counties
that are non-adjacent to metropolitan and
either non-adjacent or adjacent to
micropolitan (with urban cluster of 10,000+
people) counties are alike.
Research Procedure…
• Identified 55 counties in NC, VA, WV, & KY that met the
rural criteria.
• Obtained data on selected county demographic and
economic characteristics from Geolytics Census data
from a university library, from the federal Census and
state websites, and from federal Bureau of Health
Professions data maintained by the ECU Center for
Health Services Research.
• Computed proportional change measures defining each
variable including decennial data from 1970 to 2000.
• Assessed variability among change measures by
assessing various principal components rotated factor
patterns.
Principal Components Analysis of County Indicators
Variable
Population Increase
Change in Long Commute
Change in Female- Headed
Households
Change in Vacant Housing
Increase in Poverty
Change in Per Capita Social
Security Income
Change in Extractive Employment
Change in Retail Employment
Change in Service Employment
Change in Government Employment
Explained Variation
a Job Loss.
b Depopulation/Economic Decline.
Rotated Factor Loadings
Factor 1a
Factor 2b
.28
.23
-.12
-.67
.48
-.54
.18
-.13
-.07
.52
.64
-.44
.65
.91
.69
.79
.37
.06
-.08
.06
27%
19%
Toward a Rural County Typology…
• Summing only the 4 indicators of employment change,
virtually all counties were negative, indicating job loss.
• Negative values underlined in factor 2 were reversed
and the underlined values were summed to indicate
demographic/economic decline.
• The distributions of continuous values of job loss and
demographic/economic decline were divided at the 33rd
and 66th percentiles and the categorical variables were
cross-tabulated to create a 9X9-cell table.
• Counties were classified according to the 2 change
measures and mapped.
Rural Transitions: Job Loss and Population Decline 1970 to 2000
In 55 Isolated* Non-Core Counties Located in Kentucky, North Carolina, Virginia, and West Virginia
) )
)
))
) ))
)) )
))
)
)
) )
) ) )
) ) ) ) ) )
) )) ) )
)
)
)
)
)
)
)
)
)
)
)
)
)
Job
Loss
1
2
3
1
)
)
)
2
)
)
)
3
) ) )
)
)
)
Population
Decline
))
)
)
) )
)
)
Job loss is declining decennial employment from 1970 to 2000
across government, retail, service, and extractive (farming, timber,
mining, and aquaculture). Population/Economic Decline is a measure
of decennial change from 1970 to 2000 in county population, job
commutes of 40+ minutes, ratio of female-headed households,
vacant housing, proportion in poverty, and proportionate change in
Social Security income.
Non-core Counties adjacent to Metropolitan Counties
Metro- and Micropolitan Core Based Counties
Defining Metropolitan and Micropolitan Statistical Areas (CBSA Method)
1. Each CBSA must contain an urban area (UA) of 10,000 or more population.
2. A Metropolitan Statistical Area has at least one UA of 50,000 or more population.
3. A Micropolitan Statistical Area has at least one urban cluster of at least 10,000 but
less than 50,000 population.
4. Outlying counties are included in a CBSA if certain commuting requirements are met.
Data Sources and Definitions: US Census Bureau and Geolytics, Inc.
Center for Health Services Research and Development
East Carolina University
Greenville, NC
*A
non-core county is considered isolated if it is not geographically adjacent
to a Metropolitan Statistical Area county.
Implications…
Rural counties vary considerably according to
Job Loss and Demographic/Economic Decline.
Such variability has consequences for the
definition and measurement of rural places
(reject null hypothesis).
Implications (continued)…
Taking variability across rural counties into
account as an important socioeconomic
and demographic ecological place
characteristic can enhance knowledge of
the effect of rural residence and a myriad
of quality of life (QOL) outcomes among
older adults.
Implications (continued)…
Taking variability across rural counties into
account facilitates targeted resource
allocation and intervention to enhance the
QOL of older residents.
Conclusions
(shame on us)…
We know little about the effect
of rural residence because of
inconsistent definitions across
federal agencies. This
inconsistency affects the utility
of data maintained by those
agencies for secondary
analysis by gerontological
researchers. We need to
advocate for a common
comprehensive definition of
rural areas that takes
ecologically-important
variability into account.
Conclusions
(shame on us again)…
Authors, reviewers, and
journal editors need to
insist upon rigorous
definition of rural
residence and
measurement description
to enhance replication
and development of a
body of knowledge.
Research on Aging Farmers
Build institutional
alliances (CDC & NIOSH
with Southern
Gerontological Society
[SGS], American Society
on Aging [ASA], and the
Gerontological Society of
America [GSA—call for
papers for 11/07 out
soon]).
Research on Aging Farmers (continued)
Encourage federal agencies (e.g., NIOSH) to include
variables pertinent to…
farming and ranching (e.g., machinery condition and
operation, farm/ranch income, seasonal hiring practices),
aging (self-reported physical and cognitive
ability/disability, medication regimen/warnings), and
injury/disease prevalence and incidence
in data gathered and maintained.
Research on Aging Farmers (continued)
A progression of research….
Descriptive (a-theoretical--such as incidence of injury among older
farmers).
Multivariate Cross-Sectional Explanatory (theoretical or a-theoretical
assessing the effects of variables intervening in the relationship
between age and injury/disability among farmers)
Multivariate Longitudinal Panel Studies Theoretically- driven
assessments of change in variables over time affecting the
probability of farm injury/disability/mortality. Enables determination
whether initial injury or illness in company with continued work ,
safety precautions, and limited cash resources lead to subsequent
injury/ disability/mortality).