Content uploaded by Tracy Schroepfer
Author content
All content in this area was uploaded by Tracy Schroepfer on Jun 20, 2014
Content may be subject to copyright.
10.1177/0164027504266560ARTICLERESEARCH ON AGINGFeld et al. / IADL CAREGIVER NETWORKS
Race/Ethnicity and Marital Status
in IADL Caregiver Networks
SHEILA FELD
RUTH E. DUNKLE
University of Michigan
TRACY SCHROEPFER
University of Wisconsin–Madison
Racial/ethnic variations in instrumental activities of daily living (IADL) caregiver
network composition were examined in a nationally representative sample of elders,
using task specificity and hierarchical compensatory theoretical perspectives. Logis-
tic regressions tested network differences among White, Black, and Mexican Ameri-
can elders (n= 531 married, n= 800 unmarried). Findings concerning racial/ethnic
differences were partially dependent on marital status, differentiation of spouses from
other informal helpers among married elders, and which racial/ethnic groups were
compared. Networks including formal caregivers did not differentiate married or
unmarried Black from White elders but were more common among unmarried Mexi-
can American elders than for comparable White and Black elders. Married Black
elders with solely informal networks were more likely than comparable White elders
to have informal helpers other than the spouse. Racial/ethnic similarities and differ-
ences in caregiver networks are discussed relative to their sociocultural context,
including marital status, elder’s and spouse’s health, and financial resources.
Keywords: caregiver networks; IADLs; race/ethnicity; marital status
During the past decade, significant strides have been made in under-
standing important issues connected to caregiver networks of people
of color. Yet, several issues remain unresolved about similarities and
differences for minority and White elders in the use of formal help and
531
AUTHORS’ NOTE: Partial support for this work was provided by National Institute of Aging
(NIA) Grant T32-AG000117 to Ruth Dunkle and a Hartford Doctoral Fellowship in Geriatric
Social Work awarded to Tracy Schroepfer. An earlierversion ofthe articlewas presente d at the 1999
annual scientific meeting of the Gerontological Society of America. We gratefully acknowl-
edge the contributions of Mary Gallant, Gloria Gibson, Carolyn Lindgren, and Linda Wray
RESEARCH ON AGING, Vol. 26 No. 5, September 2004 531-558
DOI: 10.1177/0164027504266560
© 2004 Sage Publications
sources of informal care (spouse, relatives, and nonrelatives). First,
despite racial/ethnic variations in marital status, there is limited
research on whether racial/ethnic differences in caregiver networks
vary for married and unmarried elders. A few studies have addressed
this issue in relation to formal services (Miller et al. 1996; Mutchler
and Bullers 1994; White-Means 2000). Only one located study inves-
tigated whether a spouse, when present, is as likely to be the only care-
giver for different racial/ethnic groups (Stoller and Cutler 1992). Typ-
ically, spouses are not distinguished from other informal helpers in
studies of formal and informal caregivers among minority and White
elders (Mutchler and Bullers 1994; Norgard and Rodgers 1997). Sec-
ond, research has focused mostly on African American and White
elders. When Hispanic caregiver networks have been compared with
those of White elders, subgroups (Puerto Rican, Cuban, and Mexican
American) often are combined (Cantor, Brennan, and Sainz 1994;
Wallace, Levy-Storms, and Ferguson 1995). Yet, Hispanic subgroups
differ in ways that may affect caregiving (Angel and Angel 1997;
Wallace, Campbell, and Chih-Yin 1994). Furthermore, studies com-
paring caregiver networks of Hispanic subgroups and African Ameri-
cans are rare, despite significant cultural and historical factors
differentiating these minority groups (Dilworth-Anderson and Burton
1999).
Furthermore, much research on race/ethnicity has not distin-
guished caregiver networks assisting with activities of daily living
(ADLs) from those helping with instrumental activities of daily living
(IADLs) (Miller et al. 1996; Mutchler and Bullers 1994; White-
Means 2000). Yet, some studies show that the impact on network
composition of race/ethnicity, and other characteristics, differs for
these types of tasks (Lubben and Becerra 1987; Mui, Choi, and Monk
1998). Differences in factors influencing who provides ADL and
IADL care are not surprising. Theoretical and empirical work indi-
cates that these two internally reliable components describe func-
tional needs (George and Fillenbaum 1985). The nature of caregiving
tasks is an important determinant of who is best suited to give care
532 RESEARCH ON AGING
during the initial development of this project and the valuable comments received from Regula
Herzog, Aloen Townsend, and anonymous reviewers on earlier versions of this article. Address
correspondence to Sheila Feld, University of Michigan,School of Social Work, 1080 South Uni-
versity, Ann Arbor, MI 48109-1106; phone: (734) 763-5971; fax: (734) 763-3372; e-mail:
sfeld@umich.edu.
(Litwak 1985) and who actually provides aid (Lo Sasso and Johnson
2002). Providing IADL and ADL care appears to affect spouses’ well-
being somewhat differently, which could influence their willingness
and ability to provide such care (Burton et al. 2003). The linkage of
adequacy of ADL aid to institutionalization confirms the significance
of identifying factors associated with receipt of ADL care (Lo Sasso
and Johnson 2002). Factors influencing who helps with IADLs are
also important. IADL tasks are essential to maintain independence in
community-dwelling elders, require significant portions of their time
each day, and such elders are more likely to need IADL than ADL aid
(Willis 1995).
In addition, some studies of caregiver networks among racial/eth-
nic groups use nonrepresentative samples (Kemper 1992; Stommel,
Given and Given 1998) and consider only a few critical influences on
network composition (i.e., nature of the caregiving task, sociodemo-
graphic characteristics, need, network characteristics, and financial
resources) (Lubben and Becerra 1987).
In this study, we address these issues using data from the nationally
representative baseline sample of the Asset and Health Dynamics
Among the Oldest Old (AHEAD) survey (Soldo et al. 1997). We
examine the composition of caregiver networks (spouse, other infor-
mal, and formal) that help elderly married and unmarried Black, Mex-
ican, and White Americans with their functional limitations.
MARITAL STATUS AND THEORETICAL
PERSPECTIVES ON CAREGIVER NETWORKS
The hierarchical compensatory and task-specific theoretical per-
spectives address who the caregivers are for community-dwelling
elders. The hierarchical compensatory and related substitution mod-
els (Cantor 1979b; Cantor and Brennan 2000; Shanas 1979) highlight
elders’ preference for care from those in intimate personal over
socially distant relationships. The task-specific model (Litwak 1985;
Messeri, Silverstein, and Litwak 1993) stresses the match between the
structure of the caregiving task (e.g., its requirements for proximal,
highly motivated, or technically skilled helpers) and the structure of
social groups such as marital dyads, children, or formal organizations.
Both perspectives make similar predictions about the typical compo-
sition of caregiver networks for instrumental and personal care needs
Feld et al. / IADL CAREGIVER NETWORKS 533
(Messeri et al. 1993). They match the well-established finding that
most of the ADL/IADL assistance community-dwelling elders
receive is from informal rather than formal sources and that informal
helpers are predominantly spouses and other kin (Spector et al. 2000).
An important difference in these perspectives concerns the circum-
stances under which those lower in the typical helper hierarchy are
included in caregiver networks. In the hierarchical compensatory
model, availability of preferred sources is critical. Helpers lower in
the preference hierarchy are expected to be used only when more pre-
ferred sources are unavailable. An available spouse is expected to be
likely to provide all needed care, although the distinctive role of the
spouse has not been tested directly (Cantor 1979a; Cantor and
Brennan 2000; Cantor et al. 1994). Absent a spouse, available chil-
dren are more likely than other relatives or friends to provide care.
Formal care is used only when informal help is not available. In the
task-specific model, availability is one of several factors affecting the
match between characteristics of the caregiving task and potential
helpers. For example, a coresiding spouse may not be the sole helper if
care needed exceeds time or energy resources of the spouse or requires
specialized skills the spouse lacks. Similarly, formal help may substi-
tute or supplement aid from proximate kin when the kin’s resources do
not meet task requirements (Messeri et al. 1993).
Prior research tends to support predictions from both perspectives
that married elders often rely solely on their spouses for assistance
with functional needs. Nevertheless, deviations from this pattern are
more in keeping with the task-specific model; other helpers supple-
ment spousal caregiving, and not all spouses are caregivers (Burton
et al. 2003; Stoller and Cutler 1992). Similarly, the predominance of
informal over formal help is consistent with both perspectives, but
considerable evidence that formal care often is used along with infor-
mal care is more consistent with the task-specific model (Mui et al.
1998).
RACIAL/ETHNIC VARIATIONS IN CAREGIVER NETWORKS
Both theoretical perspectives suggest that similar basic principles,
based on hierarchical preferences or matching, affect network compo-
sition in various racial/ethnic groups. They also recognize the impact
on network composition of racial/ethnic variations in cultural values,
534 RESEARCH ON AGING
which could affect preferences of elders or proximity and commit-
ment of social network members, and in social structures or resources,
which could affect availability and match of the structure of social
groups to tasks (Cantor et al. 1994; Litwak 1985). Other theorists
stress that values about appropriate sources of elder care are not only
tied to the traditional cultures of countries of origin but also are linked
inextricably to current and historical social positions that affect need
for care and resources (Angel and Angel 1997; Dilworth-Anderson,
Williams, and Cooper 1999). Social location and minority/majority
group status can influence potential informal and formal caregivers’
availability and their social structures and resources. They also influ-
ence cultural norms and practices, health, access to health care, and
social and economic resources, all of which form the sociocultural
context for caregiving.
Examination of this context for care suggests that minority elders
might be less likely than are White elders with similar functional limi-
tations to use community-based formal services, thereby possibly
increasing demands for informal caregiving. Suggested barriers to use
of formal services by minority elders include financial constraints,
limited geographic availability, discrimination by service providers, a
history of insensitive or poor quality services, and cultural values
(Mui et al. 1998). The hierarchical and task-specific models suggest
that availability barriers could decrease reliance on formal help. The
hierarchical model also points to the influence of barriers on minority
elders’ preferences. The task-specific model highlights the relevance
of such barriers to the match between the commitment and skills of
formal helpers and minority elders’ needs.
The sociocultural context for care among minority and White
elders also suggests that there is a greater diversity of available and
acceptable informal helpers for minority than for White elders
(Dilworth-Anderson and Burton 1999), which may foster reliance on
informal rather than formal helpers (Angel and Angel 1997). The
racial/ethnic context for caregiving may also reflect the communal/
collectivist orientations to mutual aid (Piercy 1998; Pyke and
Bengtson 1996) and extended family structures said to characterize
African American and Mexican American families (Dilworth-Ander-
son and Burton 1999). This could reduce sole reliance on spousal
caregiving among minority compared with White elders. Some evi-
dence indicates that communal caregiving is quite common among
Feld et al. / IADL CAREGIVER NETWORKS 535
African American families (Dilworth-Anderson et al. 1999), perhaps
more so than in White families (Stommel et al. 1998), and that a wider
set of individuals is included in caregiver networks of Black than
White Americans (Dilworth-Anderson and Burton 1999), although
none of these studies focused on married elders and spousal
caregiving. The familism of Mexican Americans is said to involve
high levels of attachment and a collectivist orientation to reciprocal
aid among members of extended families (Aranda and Knight 1997).
Thus, there are various reasons for suspecting that spouses in minority
families may be less likely than are those in White families to be solo
caregivers.
The preponderance of studies based on multivariate analyses of
representative local or national samples of community-dwelling
elders, however, has not supported these expectations about differ-
ences in the caregiver networks of Black and White elders. These
studies have not typically found that Black elders are less likely to use
formal care or are more likely to rely on informal caregivers than com-
parable White elders (Miller et al. 1996; Mutchler and Bullers 1994;
Netzer et al. 1997; Norgard and Rodgers 1997; Peek, Coward, and
Peek 2000; Tennstedt, Chang, and Delgado 1998; White-Means
2000), although there are some exceptions (Wallace et al. 1998;
Wolinsky and Johnson 1991). Most of these studies combined ADL
and IADL help and did not consider the impact of marital status on
this conclusion. Three studies found no race differences in the use of
formal care by either married or unmarried elders (Miller et al. 1996;
Mutchler and Bullers 1994; White-Means 2000). In a sample
restricted to elders living only with their spouses, Stoller and Cutler
(1992) did not find race differences in sole reliance on spousal care-
givers for help with ADLs or with IADLs, nor did Black and White
elders differ in whether their non-spousal caregivers for either type of
assistance included any formal helpers. Similarly, Pezzin and
Schone’s (1999) study of unmarried Black and White elderly parents
receiving ADL/IADL care did not find racial differences in care from
children or formal sources. However, a synthesis of data from several
convenience samples found that Black primary caregivers providing
IADL and household care were more likely to have supplemental
caregivers than were White ones, but did not find race differences in
ADL care (Stommel et al. 1998).
536 RESEARCH ON AGING
There is a dearth of evidence comparing caregiver networks of rep-
resentative samples of Mexican American elders with those of Black
or White elders. The one such identified multivariate study was lim-
ited to Medicaid recipients in California and controlled only marital
status and age (Lubben and Becerra 1987). For certain IADL tasks,
Mexican American elders were less likely to get any formal help and
were more likely to receive any informal help than were comparable
White elders; differences were not found for other IADL tasks or ADL
tasks.
Important methodological limitations and variations limit the
robustness of the accumulating evidence for considerable racial/eth-
nic similarities in caregiver network composition. Most studies com-
bine ADL and IADL assistance. Studies of married elders often com-
bine spousal with other informal helpers. Sampling inclusion criteria
often differ (e.g., all elders, only those with functional limitations,
only married elders in two-person households, or only unmarried
elders with adult children). Definitions frequently vary for what con-
stitutes formal help (e.g., specific service programs or paid help with
specific tasks) and informal help (e.g., all unpaid helpers, only kin, or
only children).
Prior research has also varied in what factors besides race/ethnicity
are controlled. Nevertheless, several important influences on care-
giver network composition have been found. Influences include the
elder’s extent and type of functional limitations; gender, age, and edu-
cation of the elder; the family network’ssize, proximity, and composi-
tion; and the amount and sources of financial resources (Katz, Kabeto,
and Langa 2000; Miller et al. 1996; Norgard and Rodgers 1997; Peek
et al. 2000; Pezzin and Schone 1999; White-Means 2000). Studies
testing the effects of these factors on network composition for married
and unmarried elders typically find they are important in both. Gen-
der, however, may only be relevant for married elders (Mutchler and
Bullers 1994). These influences on network composition often reflect
racial/ethnic variations in the sociocultural context for caregiving that
might affect elders’ preferences for caregivers, caregiver availability,
or the match between characteristics of caregivers and caregiving
tasks, issues raised in Cantor’s (1979b) and Litwak’s (1985) theoreti-
cal perspectives. Spouses’ health influences married elders’ caregiver
network composition (Katz et al. 2000), but its impact does not appear
to have been controlled in studies of racial/ethnic network differences.
Feld et al. / IADL CAREGIVER NETWORKS 537
THE PRESENT STUDY
We investigated racial/ethnic differences in the inclusion of formal
and informal helpers in caregiver networks among representative
samples of 531 married and 800 unmarried elders receiving assistance
with IADL needs. We differentiated spousal, other informal, and for-
mal helpers among the married and formal and informal helpers
among the unmarried Black, Mexican American, and White elders.
Sociodemographic factors, need for help, family availability, spousal
impairment, and financial resources were taken into account.
Although it would have been desirable to study racial/ethnic differ-
ences in both ADL and IADL caregiver networks, the low numbers of
Black and Mexican American elders receiving ADL assistance pre-
cluded doing this in the present study.1
Specifically, we addressed two questions about differences in the
IADL caregiver networks of elderly African Americans, Mexican
Americans, and White Americans.
1. Are there racial/ethnic variations in the extent to which elders with a
coresident spouse/partner rely solely on their spouses, use other infor-
mal sources of aid, or use any formal help to assist with IADL tasks?
2. Are there racial/ethnic variations in the extent to which unmarried
elders rely solely on informal sources of aid, or use any formal help to
assist with IADL tasks?
Method
SAMPLE
The data were from the first wave of the AHEAD survey, conducted
in 1993-1994 (see Soldo et al. 1997 for sampling design details). This
survey had a complex multistage design. It yielded a nationally repre-
sentative sample of more than 8,000 community-dwelling elders,
aged 70 and older at the time of the initial interview, and their spouses
or partners.
The study was based on two subsamples: married respondents
(those living with a spouse or partner) and unmarried respondents
(anyone not currently living with a spouse or partner), among whom
538 RESEARCH ON AGING
84% to 89% were widowed in each racial/ethnic group. Only persons
self-identifying as non-Hispanic White, non-Hispanic Black or non-
Hispanic African American, or as Mexican American or Chicano
were included. Additional inclusion criteria were age 70 and older,
had a limitation in one or more of four IADLS, and received help with
one or more IADL limitations at least once a week. The IADL tasks
were preparing a hot meal, shopping for groceries, making a tele-
phone call, and taking medications. Questions about each IADL in the
AHEAD interview enabled us to identify limitations by reports of not
being able to perform a task without help or of not doing a task for
health reasons. Elders who performed a task without help were not
asked if they had difficulty with the task, preventing identification of
elders who may have needed help but did not receive it. Data on pay-
ment to helpers were obtained only from those getting help at least
weekly, necessitating restriction of analyses to those elders.2
The final subsamples resulted from the sample selection criteria
and deletion of influential outliers identified by regression diagnostic
procedures. They included 531 married3(404 Whites, 83 Blacks, and
44 Mexican Americans) and 800 unmarried (585 Whites, 168 Blacks,
and 47 Mexican Americans) elders. We included Mexican Americans
despite their small numbers because of the lack of prior research using
appropriate multivariate controls and representative samples that has
compared their caregiver networks with those of White or Black
Americans. We view any observed differences between networks of
Mexican Americans and other racial/ethnic groups as primarily useful
for providing valuable directions for future research, while recogniz-
ing that the absence of differences may result from power limitations
related to sample size.
DEPENDENT VARIABLE
The caregiver network variable was based on replies to two ques-
tions: (1) Who helps the most with the set of four IADL tasks? and (2)
Who else helps most often? We categorized replies to reflect predic-
tions from the hierarchical compensatory and task-specific models of
the predominance of informal over formal caregivers and the hierar-
chical compensatory model’s assumptions about the distinctive role
of spouses and the use of formal helpers in the absence of informal
Feld et al. / IADL CAREGIVER NETWORKS 539
sources. For all help recipients, we differentiated networks with any
formal helper from those with only informal helpers. For the married,
spouses and other informal helpers were further differentiated yield-
ing three types of networks: any formal, denoting at least one paid or
organizational helper, regardless of the presence or type of any second
helper (e.g., another formal helper, spouse/partner, unpaid helper
other than spouse/partner); spouse only, indicating only the coresident
spouse/partner provided help; and informal, indicating at least one
unpaid helper other than the spouse/partner (although the spouse/part-
ner might have also provided help) and no paid or organizational help.
For the unmarried, any formal and informal networks were differenti-
ated. As noted, any formal networks could consist of only formal
helpers (as they did for 36% of the married and 72% of the unmarried).
Any formal networks could also include a formal and informal helper,
as they did for 64% of the married (including 38% with a spouse and
26% with another informal helper) and for 28% of the unmarried
respondents. Informal networks of married elders could include a
spouse and nonspousal helper (43%) or only nonspousal help (57%).
INDEPENDENT VARIABLES
We describe the independent variables in Table 1. Race/ethnicity
was the focal variable, coded as White, Black, or Mexican American.
To address the sociocultural context for caregiving, we included four
types of variables that prior theory and research indicate are likely to
be associated with caregiver network composition and race/ethnicity.
Demographic variables were gender, age, and years of education.
Need for help variables included three indicators of health: number of
IADL limitations, number of ADL limitations, and the presence of
serious cognitive problems. Family network variables assessed poten-
tial availability of family helpers. Number of proximate children
(those coresiding or living less than 10 miles away) and number of rel-
atives (siblings and grandchildren) were used for both the married and
unmarried. For married elders, availability was also indexed by two
measures of spousal health: Spouse did not have IADL limitations,
and spouse did not have ADL limitations. Medicaid and the poverty
ratio measured financial resources that could facilitate use of formal
help.
540 RESEARCH ON AGING
541
TABLE 1
Measures of Independent Variables
Variable Description (Range)
Race/ethnicity White: non-Hispanic White
Black: non-Hispanic Black or African American
Mexican American: Mexican American or Chicano
Demographics
Gender 0 = female
1 = male
Age Age in years (70-103)
Education Years of schooling, truncated at 17 (0-17)
Need for help
Number of IADLs IADL limitations were not being able to prepare a hot meal, to shop for groceries, to make a telephone call, or to
take medication without help; or not doing the task because of health reasons (1-4)
Number of ADLs ADL limitations were elder gets help, does not do, or has difficulty in dressing, bathing, eating, or toileting; or elder
gets help, does not do, uses equipment, or has difficulty walking or getting in and out of bed (0-6)
Cognitive problems 0 = no serious cognitive problems
1 = elder gave less than half correct replies to phone version of Mini Mental Status Examination, or proxy indicated
behavior symptomatic of cognitive problems (wandering, getting lost, hearing voices, or unable to leave alone)
Family network
Number of proximate children Number of children of elder and spouse/partner coresiding or living less than 10 miles away from parent (0-8)
Number of relatives Number of living siblings and grandchildren of elder and spouse/partner truncated at 19 (0-19)
Spouse no IADLs 0 = spouse has any IADL limitations, as previously defined in number of IADL limitations
1 = spouse has no IADL limitations, as previously defined in number of IADL limitations
Spouse no ADLs 0 = spouse has any ADL limitations, as previously defined in number of ADL limitations
1 = spouse has no ADL limitations, as previously defined in number of ADL limitations
Financial resources
(continued)
542
Medicaid 0 = no Medicaid coverage
1 = self-report that health care currently covered by Medicaid
Poverty ratio Ratio of prior year’s household income to U.S. poverty figure for that year (1992/1993) for a given household size
and composition truncated at 13.0 (0.0-13.0)
NOTE: ADL = activity of daily living; IADL = instrumental activity of daily living.
TABLE 1 (continued)
Variable Description (Range)
ANALYTIC STRATEGY
To obtain appropriate descriptive statistics and standard errors, svy
commands in Stata’s statistical packagewere used to take into account
AHEAD’s complex multistage sample design (StataCorp 1999:322-
24). These commands adjust for sampling weights, clustering, and
stratification of the sample by geographic location and size of place.
Normalized weights were used to adjust for sample selection proba-
bilities (oversampling of minority elders, nonresponse, and deviations
from the 1990 census). The commands also take into account multi-
stage sampling, that is, clustering of the sample within primary and
secondary sampling units and implicitly for clustering within house-
holds (interviewing both members of couples); “They allow any
amount of clustering within the primary sampling unit . . . [and] pro-
duce variance estimates that generally will be either approximately
unbiased, or, if biased, they will be toward more conservative esti-
mates” (StataCorp 1999:324).
The effects of race/ethnicity, and other independent variables (12
for the married and 10 for the unmarried), on caregiver network com-
position were tested in separate logistic regression models for the two
subsamples. Stata’s svy procedures yield Fvalues and significance
levels for the total model, but no other goodness-of-fit statistics.
For the married subsample, multinomial logistic regression was
used, and the model including the 13 independent variables was run
twice in order to obtain all comparisons among the three types of net-
works comprising the dependent variable. In each run of the model,
we used two dummy variables to represent the three network types.
One run compared elders who relied on either of the two types of
informal networks (spouse only and informal) with those with formal
networks (the omitted referent category of any formal). In the second
run of the model, the reference category was informal, yielding a new
comparison between the two types of informal networks (spouse only
versus informal). The comparison between any formal and informal
networks in the second run replicated the same comparison from the
first run of the model (with signs reversed); for this reason, only the
results from the first run are shown in tabular form.
Binomial logistic regression was used for the unmarried elders to
identify the effects on network composition of race/ethnicity and the
Feld et al. / IADL CAREGIVER NETWORKS 543
other 10 predictors. Informal networks were compared with any for-
mal networks, the omitted reference category.
In the logistic regression models described above for both sub-
samples, the models were run twice to obtain comparisons among all
three of the racial/ethnic groups. In all runs of the models, race/ethnicity
was entered as two dummy variables, but the reference category dif-
fered. In one run, the omitted reference category was White elders,
yielding comparisons of the Black and Mexican American elders’ net-
works with those of Whites. The reference category in the other run
was Black elders, yielding a new comparison of the networks of Mexi-
can Americans with those of Black elders. The comparison between
White and Black elders’networks from this run replicated data for the
same comparison from the first run (with signs reversed) of the mod-
els; for this reason, only the results from the first run of each model are
shown in tabular form.
Results
DESCRIPTIVE RACIAL/ETHNIC DIFFERENCES
The data in Table 2 describe variations in the IADL caregiver net-
works and the sociocultural context of caregiving for the Black, Mexi-
can American, and White elders. Significant racial/ethnic differences
in network composition were found only for married respondents.
The most common pattern for married White and Black Americans
was a network including only the spouse, but Mexican Americans’
networks were as likely to consist solely of the spouse as to include
other informal helpers. Networks including a nonspousal informal
helper were more common among married Mexican American and
Black elders than White elders. Networks including any formal helper
were least common in all three married groups, but such networks
were most likely among Mexican American elders. For unmarried
elders in all racial/ethnic groups, solely informal networks were much
more common than were those with any formal helper. Most other
variables showed significant racial/ethnic differences for both sub-
samples. These differences were taken into account in the multivariate
analyses of network composition.
544 RESEARCH ON AGING
545
TABLE 2
Racial/Ethnic Differences Within Marital Status
Married Unmarried
White Black Mexican White Black Mexican
Variable (n= 404) (n= 83) (n= 44) p(n= 585) (n= 168) (n= 47) p
Dependent variable
Caregiver network *
Spouse only 77.7% 55.4% 36.4%
Informal 15.6% 37.3% 36.4% 71.6% 79.8% 63.8%
Any formal 6.7% 7.2% 27.3% 28.4% 20.2% 36.2%
Independent variables
Gender
Female 45.8% 38.6% 40.9% 86.3% 86.9% 78.7%
Male 54.2% 61.4% 59.1% 13.7% 13.1% 21.3%
Mean age 79.1 79.2 79.5 83.5a80.5b81.1b*
Mean education 10.4a7.1b3.5c*9.9
a7.6b3.2c*
Mean number of IADLs 1.9a2.2b2.0a,b *1.8
a2.1b1.8a,b *
Mean number of ADLs 2.2 2.5 2.3 2.4 2.6 2.5
Cognitive problems * *
No 79.2% 54.2% 75.0% 75.6% 58.3% 57.4%
Yes 20.8% 45.8% 25.0% 24.4% 41.7% 42.6%
Mean number of proximate children 1.0a1.2a2.8b*1.1
a1.5b2.2c*
Mean number of relatives 7.7a7.9a14.4b*7.2
a8.5b14.0c*
Spouse no IADLs *
Has IADLs 18.1% 34.9% 68.2%
No IADLs 81.9% 65.1% 31.8%
(continued)
546
Spouse no ADLs *
Has ADLs 28.0% 32.5% 47.7%
No ADLs 72.0% 67.5% 52.3%
Medicaid **
No 93.3% 78.3% 50.0% 79.8% 56.5% 42.6%
Yes 6.7% 21.7% 50.0% 20.2% 43.5% 57.4%
Mean poverty ratio 2.7a1.6b1.1c*2.3
a1.8b1.6b*
NOTE: Means and percentages are based on raw data. All significance valuesare for two-tailed tests and take design effects into account by using the Stata software.
For categorical variables, the Pearson chi-square with the Rao and Scott second-order correction was used (svytab); for continuous variables, regression procedures
were used (svyreg), as there is no Stata procedure analogous to analysis of variance. Means in the same row that have differentsuperscripts differ at p≤.05. ADL =
activity of daily living; IADL = instrumental activity of daily living.
*p≤.05.
TABLE 2 (continued)
Married Unmarried
White Black Mexican White Black Mexican
Variable (n= 404) (n= 83) (n= 44) p(n= 585) (n= 168) (n= 47) p
MULTIVARIATE RACIAL/ETHNIC DIFFERENCES
Table 3 shows the findings from logistic regressions addressing our
research questions about racial/ethnic variations in caregiver network
composition. We report unstandardized (b) coefficients to show sig-
nificance levels; associated odds ratios show how membership in spe-
cific racial/ethnic groups affected the odds of having a particular care-
giver network pattern. Standardized (β) coefficients are reported to
compare the relative impact of race/ethnicity and the other independ-
ent variables on network composition (Selvin 1996). We derived these
coefficients by dividing each unstandardized coefficient by its esti-
mated standard error, as these estimates take clustering of the sample
into account; these coefficients can be greater than +1 or –1.
Married elders. The first research question asked whether IADL
caregiver networks of married Black, Mexican American, and White
elders differ when other correlates of network composition are taken
into account. Data from the two multinomial logistic regression runs
presented in the first three columns of Table 3 indicate that the overall
answer to this question is a partial yes. Married Black elders’ networks
differed from White and Mexican American elders in their inclusion
of informal helpers other than the spouse, but married White and Mex-
ican American elders’ networks did not differ.
The first two columns of Table 3 present findings from the model
run with networks including any formal helper as the reference cate-
gory to which the two types of informal caregiver networks were com-
pared. As data in the first column show, the racial/ethnic groups did
not differ in reliance on spouse only and any formal networks. Other
variables did significantly differentiate elders from these two types of
networks. The magnitude of the standardized regression coefficients
indicate that the elder’s number of IADL limitations had the greatest
relative impact (β= –4.42), with more limitations decreasing sole reli-
ance on the spouse versus using any formal help, as did Medicaid cov-
erage (β= –3.21). Being a male help recipient (β= 3.05) and having a
spouse without IADL limitations (β= 2.94) increased the likelihood
of solo spousal networks.
The second column of Table 3 shows one significant racial/ethnic
difference when married elders whose network included an informal
helper other than (or in addition to) the spouse were compared with
Feld et al. / IADL CAREGIVER NETWORKS 547
548 RESEARCH ON AGING
TABLE 3
Logistic Regression Models of Characteristics
Predicting the Composition of Caregiver Networks
for Married (n= 531) and Unmarried (n= 800) Elders
Married Unmarried
Spouse Only/ Informal/ Spouse Only/ Informal/
Any FormalaAny FormalaInformalbAny Formal c
bdbdbdbd
βeβeβeβe
Variable Odds Ratio Odds Ratio Odds Ratio Odds Ratio
Black/Whitef–0.09 0.86 -0.96** 0.44
–0.19 1.34 -2.69 1.59
0.91 2.36 0.38 1.55
Mexican/Whitef–0.79 –1.15 0.36 –1.35*
–1.33 –1.40 0.46 -2.33
0.45 0.32 1.43 0.26
Mexican/Blackg–0.70 –2.01* 1.32† –1.79***
–0.94 –2.25 1.69 -3.58
0.50 0.13 3.74 0.17
Male 1.27** 0.39 0.88** –0.11
3.05 0.94 2.84 -0.45
3.56 1.48 2.41 0.90
Age –0.04 0.01 –0.05† –0.04**
–1.05 0.15 –1.88 –2.88
0.96 1.01 0.95 0.96
Education 0.04 –0.03 0.07 -0.06*
0.70 –0.40 1.23 –2.15
1.04 0.97 1.07 0.94
Number of IADLs –0.68*** –0.42† –0.26 0.04
–4.42 –1.97 –1.53 0.46
0.51 0.66 0.77 1.04
Number of ADLs –0.22† –0.03 –0.19* –0.23***
–1.96 –0.31 –2.20 –4.79
0.80 0.97 0.83 0.80
Cognitive problems 0.24 –0.64 0.88* –0.21
0.47 –1.36 2.27 –1.11
1.27 0.53 2.41 0.81
Number of proximate children –0.24 0.06 –0.30** 0.51***
–1.29 0.33 –3.01 4.92
0.79 1.06 0.74 1.67
Number of relatives 0.06 0.05† 0.01 0.03
1.63 1.73 0.27 1.43
1.06 1.05 1.01 1.03
Spouse no IADLs 1.75** –0.64 2.39***
2.94 -1.12 6.87
5.75 0.53 10.91
those whose networks included any formal helper. The networks of
Black and Mexican American elders did not differ from those of
White elders, but the model run with Blacks as the reference category
showed a difference in the Mexican American and Black elders’ net-
works. The odds ratio (0.13) shows married Mexican American elders
were 87% less likely than were comparable Black elders to use an
informal helper other than the spouse than to use any formal help.
Both measures of financial resources significantly reduced reliance
on informal helpers other than the spouse compared to using formal
Feld et al. / IADL CAREGIVER NETWORKS 549
Spouse no ADLs 1.65** 0.67 0.98**
2.74 1.39 2.88
5.21 1.95 2.66
Medicaid –1.62** –1.44* –0.18 –0.69**
–3.21 –2.50 –0.31 –3.22
0.20 0.24 0.84 0.50
Poverty ratio –0.14 –0.34** 0.20† 0.00
–1.63 –2.66 1.77 0.05
0.87 0.71 1.22 1.00
NOTE: All significance levels are for two-tailed tests. ADL = activity of daily living; IADL = in-
strumental activity of daily living.
†p≤.10. *p≤.05. **p≤.01. ***p≤.001.
a. Results from multinomial logistic model run with “any formal” as the reference category.
b. Results from the multinomial logistic model run with “informal” as the reference category.
Figures are only shown for the comparison between “spouse only” and “informal”help, as those
for the comparison between “any formal” and “informal” help are identical (with signs reversed)
to those presented in column 2. The overall Fvalue for both runs was 14.61, p= .0000.
c. Results from the binomial logistic regression model run with “any formal” help as the refer-
ence category. The overall Fvalue was 6.77, p= .0000.
d. Unstandardized regression coefficients.
e. Standardized regression coefficients.
f. Results from models run with “White” as the reference category.
g. Results from models run with “Black’as the reference category. Figures are shown onlyfor the
comparison between Mexican American and Black elders, as those for the comparison between
White and Black elders are identical (with signs reversed) to those presented above for the com-
parison between Black and White elders’ networks.
TABLE 3 (continued)
Married Unmarried
Spouse Only/ Informal/ Spouse Only/ Informal/
Any FormalaAny FormalaInformalbAny Formal c
bdbdbdbd
βeβeβeβe
Variable Odds Ratio Odds Ratio Odds Ratio Odds Ratio
help. The standardized coefficients for these measures (β= –2.66 for
the poverty ratio and β= –2.50 for Medicaid) indicate they had a
somewhat larger impact than being Mexican American rather than
Black (β= –2.25).
The third column of Table 3 presents the findings comparing solo
spousal networks with those including nonspousal informal helpers,
obtained from the run of the multinomial model in which informal
networks were the reference category. It shows that the informal help-
ers of married elders who did not use any formal aid differed for Black
and White elders, but not for Mexican American and White elders.
The odds ratio (0.38) shows that Black elders were 62% less likely
than comparable White elders to rely solely on their spouses than to
use any other informal helper. Networks of Mexican American and
Black elders who relied on informal helpers also tended to differ (p=
.098). The similarity of this difference to that between Black and
White elders became clear when Mexican Americans were made the
reference category compared to Black elders (data not shown). This
reversed odds ratio (0.27) indicates that these Black elders were 87%
less likely than were comparable Mexican American elders to rely
solely on their spouses than to use an informal helper other than the
spouse.
The standardized regression coefficients indicate that availability
of primary family and being male had greater relative impacts on
informal network composition than did being Black or White.
Increases in the odds of a solo spousal network, compared with using
other informal helpers, were associated with a spouse without any
IADL limitations (β= 6.87), fewer proximate children (β= –3.01), a
spouse without ADL limitations (β= 2.88), male help recipients (β=
2.84), and with being White rather than Black (β= –2.69).
Unmarried elders. The second research question asked whether
IADL caregiver networks of unmarried Black, Mexican American,
and White elders differ when other correlates of network composition
are taken into account. Data from the binomial logistic regression (in
the last column of Table 3) indicate that networks of unmarried Mexi-
can Americans differed from those of Black and White elders, but
those of Black and White elders did not differ. Unmarried Mexican
American elders’ networks were significantly less likely than were
those of comparable White and Black elders to contain only informal
550 RESEARCH ON AGING
helpers than to include formal help. The odds ratios (0.26 and 0.17,
respectively) show that solely informal networks were 74% less likely
to occur for unmarried Mexican American than for comparable White
elders and were 83% less likely to occur for Mexican American than
Black elders.
The standardized regression coefficients for other significant vari-
ables show that having more proximate children (β= 4.92) and fewer
help recipient ADL limitations (β= –4.79) had the strongest relative
impacts on the odds of unmarried elders having solely informal net-
works versus networks including formal help. The effect of being
Mexican American or Black was weaker (β= –3.58) than that of these
variables, but somewhat stronger than that of the other significant
variables (Medicaid coverage, β= –3.22; age, β= –2.88; being Mexi-
can American or White, β= –2.33; and education, β= –2.15).
Discussion
The goal of this study was to advance our understanding of racial/
ethnic variations in caregiver network composition. Using data from a
nationally representative sample, we considered many factors associ-
ated with network composition, focused on IADL tasks, and drew on
the hierarchical compensatory and task-specific perspectives on net-
work composition to achieve this end. We addressed three issues that
had not received sufficient attention: (1) whether racial/ethnic varia-
tions in networks were similar for married and unmarried elders; (2)
whether distinguishing spousal and other informal helpers among
married elderly influenced conclusions about racial/ethnic variations
in network composition and (3) whether networks of comparable
Mexican American, White, and Black American elders varied. Find-
ings indicate that conclusions about racial/ethnic differences in IADL
caregiver networks depend in part on the marital status of the elders
and on differentiating spouses from other informal helpers of married
elders, as suggested by both theoretical models. Results also indicate
that Black and White Americans differ only in the composition of
their informal caregiver networks. Findings also suggest that the
inclusion of formal helpers in caregiver networks may differ for
elderly Mexican Americans and comparable White and Black
Americans.
Feld et al. / IADL CAREGIVER NETWORKS 551
Overall, the findings confirm and extend those from prior research
on the inclusion of formal helpers in the caregiver networks of elderly
Black and White Americans and are in keeping with similar predic-
tions from both theoretical models about the low likelihood of reli-
ance on formal help. In representative community samples, reliance
on formal help to meet various long-term care needs is uncommon and
typically has not been found to vary by race in multivariate analyses
with appropriate controls. This study confirmed these conclusions for
both married and unmarried elders: Use of formal IADL help was
rare, and Black and White Americans’reliance on formal help did not
differ when other major influences on network composition were con-
sidered. The present study also enhanced the robustness of this con-
clusion for married elders. The use of formal help was similar for mar-
ried Black and White elders even when (1) racial differences in the
spouse’s impairments in instrumental and basic activities of daily liv-
ing were taken into account, and (2) networks including formal help
were compared with informal networks composed solely of spouses
or those including informal helpers other than the spouse.
Our findings did show, however, that Black elders with solely infor-
mal networks were significantly more likely than were comparable
White elders to have informal helpers other than the spouse than to
rely solely on their spouses. As more than 90% of married Black and
White elders only had informal IADL caregivers, this difference is
noteworthy. Our finding is consistent with portrayals of Black fami-
lies having a communal rather than individualistic orientation to
caregiving responsibilities (Dilworth-Anderson et al. 1999). It is also
congruent with evidence that Black families are more likely than
White Americans to endorse the sharing of caregiving responsibilities
(Dilworth-Anderson and Burton 1999; Lee, Peek, and Coward 1998;
Stommel et al. 1998). Yet, Stoller and Cutler’s (1992) multivariate
study of a representative sample of married elders did not find a racial
difference in sole reliance on the spouse. This inconsistency with our
finding may be related to the limitation of that study’s sample to mar-
ried couples living in two-person households and the lack of control
for spousal impairment. We found no indication, however, that mar-
ried Mexican Americans were more likely to have communal infor-
mal networks than were comparable White elders. Further study of
Mexican American networks is warranted given descriptions of Mexi-
can American familism as involving a collectivist orientation (Aranda
552 RESEARCH ON AGING
and Knight 1997) and the possibility that the sample size of Mexican
Americans in the present study was insufficient to detecta difference.
The hierarchical compensatory and task-specific models help us
understand these findings about married elders’ informal caregiver
networks. Both perspectives predict high reliance on the spouse. The
hierarchical model assumes spouses are the most preferred helpers
and, when available, the spouse is likely to provide all the needed care.
The task-specific model proposes that the structures of marital dyads
and of tasks related to functional limitations are highly likely to
match. Consistent with both models, solo spousal IADL caregiver
networks were most prevalent for White and Black elders. The finding
that sole reliance on the spouse was more common for White than
Black elders (77.7% vs. 55.4%), a difference sustained in the multi-
variate analyses for elders whose networks consisted entirely of infor-
mal helpers, appears more consistent with the task-specific model.
This model recognizes that characteristics of informal groups besides
the marital dyad may match the structure of some tasks. It also
acknowledges racial/ethnic variations in informal group structures
that could affect the likelihood of this match.
We view the present findings about the use of formal IADL care-
givers by Mexican American elders with considerable caution
because they are based on small samples. Nevertheless, noting them
seems useful, as little is known about this growing segment of our
elderly population. Unlike prior research, the findings stem from
multivariate analyses that considered relevant variables and are based
on respondents from a representative sample of Americans. The sig-
nificant differences showed a greater likelihood of inclusion of formal
helpers in the networks of Mexican American elders than of compara-
ble White or Black elders. These unexpected findings clearly need
replication. Such work would benefit from Dilworth-Anderson’s and
Burton’s (1999) recommendation for a more detailed analysis of the
similarities and differences in the sociocultural context of various
minority groups. For example, it is often assumed that Mexican and
Black Americans face similar discrimination in the health and social
service systems (Lubben and Becerra 1987; Wallace et al. 1995). Yet,
recent evidence indicates that Black Americans are more likely than
are Hispanic Americans to hold strong beliefs about racial discrimina-
tion in the health care system (La Veist, Chamberlain, and Jarrett
2000) and to experience stress due to racial bias (Williams 2000). It is
Feld et al. / IADL CAREGIVER NETWORKS 553
also possible that Mexican American families absorb paid personal
care workers into their family systems in different ways than other
families do (Aranda and Knight 1997).
Our findings also indicate that health and financial resources typi-
cally had stronger relative net impacts on network composition than
did race/ethnicity. Health resources fostered sole reliance on the
spouse. Spouses without IADL limitations were much more likely to
be solo caregivers than were spouses with such limitations. The
spouse’s ADL limitations and the number of the elder’s IADL and
ADL limitations had similar, though smaller, effects. We view these
findings as consistent with the task-specific model. These are situa-
tions where the health resources of the spouse are likely to match the
task needs of the care recipient, either because the spouse has the func-
tional abilities needed to provide care or the care recipient has a level
of functional limitations that requires limited energy or skills from the
spouse. Personal and governmental financial resources fostered the
inclusion of formal helpers in caregiver networks. We found these
effects despite controls for availability of informal helpers, suggesting
that the generally low use of formal helpers does not simply reflect a
hierarchy of preferences but also the inaccessibility of formal ser-
vices. Racial/ethnic differences in health and financial resources are
part of the sociocultural context for caregiving. This indicates the
importance of considering the functional health of the marital pair in
research on racial/ethnic variations in spousal caregiving and the
health and financial resources of married and unmarried elders in
studies of reliance on formal caregivers.
Despite the usefulness of these findings, we recognize several limi-
tations in the study. Beyond the need to replicate the findings concern-
ing the Mexican American elders due to their very small sample size,
the size of the sample of married Black elders may have limited the
statistical power of racial comparisons. These numbers also limited
the kinds of variations in the composition of informal networks we
could study and precluded analyses of possible differences in the
types of formal helpers used for specific IADL tasks. Limitations in
the interview questions prevented us from investigating racial/ethnic
differences in networks used for specific IADL tasks or those among
elders who had IADL difficulties but received no help or received help
less than weekly. We were unable to determine if there were differ-
ences in the impact of race/ethnicity, or other factors, on IADL and
554 RESEARCH ON AGING
ADL networks because the available sample of elders receiving ADL
help was too small.
Nonetheless, this study advances knowledge about community-
based networks that assist minority and White elders with IADL diffi-
culties and suggests significant directions for practitioners and policy
makers. Findings indicate that informal caregiver networks of married
Black elders are more likely to include informal helpers other than the
spouse than are those of comparable White elders. This suggests prac-
titioners may be able to involve a greater range of supporters to assist
married Black than White elders. Health of both the focal elder and his
or her spouse influenced network composition, indicating that the
health of both members of elderly couples is important to consider
when practitioners assess the need for additional informal or formal
help and policy makers set program eligibility criteria. The data also
show that minority elders are at least as willing as comparable White
elders are to rely on formal help, suggesting practitioners should not
hesitate to refer minority elders to formal help resources.
NOTES
1. There would have been 244 married and 326 unmarried cases with data on activities of
daily living (ADL) networks, using the same selection criteria noted in the Sample section. These
married cases included 183 White, 42 Black, and 19 Mexican American elders; the unmarried
numbers were 238, 72, and 16, respectively.
2. This resulted in the loss of approximately 6% of otherwise eligible married elders and 10%
of the unmarried subgroup. Logistic regressions (not shown) were run to see if there were racial/
ethnic differences in the probability of elimination for this reason, controlling all other variables
(described in the Independent Variables section). The one significant difference showed that
unmarried Black elders were more likely than comparable White elders to be eliminated due to
receiving help less than weekly (p= .001).
3. In 52 couples, both partners met all inclusion criteria and were both included in the sample
of 531 married respondents. The Analytic Strategy section describes how appropriate standard
errors were obtained, given this clustering within households.
REFERENCES
Angel, Ronald J. and Jacqueline L. Angel. 1997. Who Will Care for Us? Aging and Long Term
Care in Multicultural America. New York: New York University Press.
Feld et al. / IADL CAREGIVER NETWORKS 555
Aranda, Maria P. and Bob G. Knight. 1997. “The Influence of Ethnicity and Culture on the Care-
giver Stress and Coping Process: A Sociocultural Review and Analysis.” The Gerontologist
37:342-54.
Burton, Lynda C., Bozena Zdaniuk, Richard Schulz, Sharon Jackson, and Calvin Hirsch. 2003.
Transitions in spousal caregivers. The Gerontologist 43:230-41.
Cantor, Marjorie H. 1979a. “The Informal Support System of New York’s Inner City Elderly: Is
Ethnicity a Factor?” Pp. 153-74 in Ethnicity and Aging, edited by D. E. Gelfand and A. J.
Kutzik. New York: Springer.
. 1979b. “Neighbors and Friends: An Overlooked Resource in the Informal Support Sys-
tem.” Research on Aging 1:434-63.
Cantor, Marjorie H. and Mark Brennan. 2000. Social Care of the Elderly: The Effects of Ethnic-
ity, Class, and Culture. New York: Springer.
Cantor, Marjorie H., Mark Brennan, and Anthony Sainz. 1994. “The Importance of Ethnicity in
the Social Support Systems of Older New Yorkers: A Longitudinal Perspective (1970-
1990).” Journal of Gerontological Social Work 22:95-128.
Dilworth-Anderson, Peggye and Linda Burton. 1999. “Critical Issues in Understanding Family
Support and Older Minorities.” Pp. 93-105 in Full Color Aging, edited by T. P. Miles. Wash-
ington, DC: Gerontological Society of America.
Dilworth-Anderson, Peggye, Sharon Williams, and Theresa Cooper. 1999. “Family Caregiving
to Elderly African Americans: Caregiver Types and Structures.” Journals of Gerontology:
Social Sciences 54B:S237-S241.
George, Linda K. and Gerda G. Fillenbaum 1985. “OARS Methodology: A Decade of Experi-
ence in Geriatric Assessment.” Journal of the American Geriatric Society 33:607-15.
Katz, Steven J., Mohammed Kabeto, and Kenneth M. Langa. 2000. “Gender Disparities in the
Receipt of Home Care for Elderly People With Disability in the United States.”Journal of the
American Medical Association 284:3022-27.
Kemper, Peter. 1992. “The Use of Formal and Informal Home Care by the Disabled Elderly.”
Health Services Research 27:421-51.
La Veist, Thomas A., Diala Chamberlain, and Nicole C. Jarrett. 2000. “Social Status and Per-
ceived Discrimination: Who Experiences Discrimination in the Health Care System, How
and Why? Pp. 194-208 in Minority Health in America, edited by C. J. R. Hogue, M. A. Har-
graves, and K. S. Collins. Baltimore, MD: Johns Hopkins University Press.
Lee, Gary R., Chuck W. Peek, and Raymond T. Coward. 1998. “Race Differences in Filial
Responsibility Expectations Among Older Parents.” Journal of Marriage and the Family
60:404-12.
Litwak, Eugene. 1985. Helping the Elderly: The Complementary Roles of Informal Networks
and Formal Systems. New York: Guilford.
Lo Sasso, Anthony T. and Richard Johnson. 2002. “Does Informal Care From Adult Children
Reduce Nursing Home Admissions for the Elderly?” Inquiry 39:279-97.
Lubben, James E. and Rosina M. Becerra. 1987. “Social Support Among Black, Mexican, and
Chinese Elderly.” Pp. 130-44 in Ethnic Dimensions of Aging,editedbyD.E.Gelfandand
C. M. Barresi. New York: Springer.
Messeri, Peter, Merrill Silverstein, and Eugene Litwak. 1993. “Choosing Optimal Support
Groups: A Reviewand Reformulation.” Journal of Health and Social Behavior 34:122-37.
Miller, Baila, Richard T. Campbell, Lucille Davis, Sylvia Furner, Aida Giachello, Thomas
Prohaska, Julie E. Kaufman, Min Li, and Carmen Perez. 1996. “Minority Use of Community
Long-Term Care Services: A Comparative Analysis.” Journals of Gerontology: Social Sci-
ences 51B:S70-S81.
Mui, Ada C., Namkee G. Choi, and Abraham Monk. 1998. Long-Term Care and Ethnicity.
Westport, CT: Greenwood.
556 RESEARCH ON AGING
Mutchler, Jan E. and Susan Bullers. 1994. “Gender Differences in Formal Care Use in Later
Life.” Research on Aging 16:235-50.
Netzer, Julie K., Raymond T. Coward, Chuck W. Peek, John C. Henretta, R. Paul Duncan, and
Molly C. Dougherty. 1997. “Race and Residence Differences in the Use of Formal Services
by Older Adults.” Research on Aging 19:300-32.
Norgard, Theresa M. and Willard L. Rodgers. 1997. “Patterns of In-Home Care Among Elderly
Black and White Americans.”Journals of Gerontology Series B 52B (Special Issue): 93-101.
Peek, M. Kristen, Raymond T. Coward, and Chuck W. Peek. 2000. “Race, Aging, and Care: Can
Differences in Family and Household Structure Account for Race Variations in Informal
Care?” Research an Aging 22:117-42.
Pezzin, Liliana E. E. and Barbara S. Schone. 1999. “Parental Marital Disruption and
Intergenerational Transfers: An Analysis of Lone Elderly Parents and Their Children.”
Demography 36:287-97.
Piercy, Kathleen W. 1998. “Theorizing About Family Caregiving: The Role of Responsibility.”
Journal of Marriage and the Family 60:109-18.
Pyke, Karen D. and Vern L. Bengtson. 1996. “Caring More or Less: Individualistic and
Collectivistic Systems of Family Eldercare.” Journal of Marriage and the Family 58:379-92.
Selvin, Steve. 1996. Statistical Analysis of Epidemiologic Data. New York: Oxford University
Press.
Shanas, Ethel. 1979. “The Family as a Social Support System in Old Age.” The Gerontologist
19:169-74.
Soldo, Beth J., Michael D. Hurd, Willard L. Rodgers, and Robert B. Wallace. 1997. “Asset and
Health Dynamics Among the Oldest Old: An Overview of the AHEAD Study.” Journals of
Gerontology, Series B 52B (Special Issue):1-20.
Spector, William D., John A. Fleishman, Liliana E. Pezzin, and Brenda C. Spillman. 2000. The
Characteristics of Long-Term Care Users. AHPQ Publication No. 00-0049. Rockville, MD:
Agency for Health Care Research and Policy.
StataCorp. 1999. Statistical Software: Release 6.0. College Station, TX: Stata Corporation.
Stoller, Eleanor P. and Stephen J. Cutler. 1992. “The Impact of Gender on Configurations ofCare
Among Elderly Couples.” Research on Aging 14:313-30.
Stommel, Manard, Charles W. Given, and Barbara A. Given. 1998. “Racial Differences in the
Division of Labor Between Primary and Secondary Caregivers.” Research on Aging 20:199-
217.
Tennstedt, Sharon L., Bei-Hung Chang, and Melvin Delgado. 1998. “Patterns of Long-Term
Care: A Comparison of Puerto Rican, African-American, and Non-Latino White Elders.”
Journal of Gerontological Social Work 30:179-99.
Wallace, Steven P., Kevin Campbell, and Chih-Yin Lew-Ting. 1994. “Structural Barriers to the
Use of Formal In-Home Services by Elderly Latinos.” Journals of Gerontology: Social Sci-
ences 49:S253-S263.
Wallace, Steven P., Lene Levy-Storms, and Linda R. Ferguson. 1995. “Access to Paid In-Home
Assistance Among Disabled Elderly People: Do Latinos Differ From Non-Latino Whites?”
American Journal of Public Health 85:970-75.
Wallace, Steven P., Lene Levy-Storms, Raynard S. Kington, and Ronald M. Andersen. 1998.
“The Persistence of Race and Ethnicity in the Use of Long-Term Care.” Journals of Geron-
tology: Social Sciences 53B:S104-S112.
White-Means, Shelley I. 2000. “Racial Patterns in Disabled Elderly Persons’ Use of Medical
Services.” Journals of Gerontology: Social Sciences 55B:S76-S89.
Williams, David R. 2000. “Race, Stress, and Mental Health.” Pp. 209-43 in Minority Health in
America, edited by C. J. R. Hogue, M. A. Hargraves, and K. S. Collins. Baltimore, MD: Johns
Hopkins University Press.
Feld et al. / IADL CAREGIVER NETWORKS 557
Willis, Sherry L. 1995. “Everyday Problem Solving.” Pp. 287-307 in Handbook of the Psychol-
ogy of Aging, 4th ed., edited by J. E. Birren and K. W.Schaie. New York:Academic Press.
Wolinsky, Federic D., and Robert J. Johnson. 1991. “The Use of Health Services by Older
Adults.” Journals of Gerontology: Social Sciences 46:S345-S357.
Sheila Feld is a professor emerita of social work at the University of Michigan. Her
current research focuses on caregiving, role transitions, dependency, and social sup-
port among the elders.
Ruth E. Dunkle is a professor of social work and codirector of the National Institute of
Aging (NIA) training program in Social Research on Applied Issues of Aging and the
Hartford Foundation initiative on geriatric social work at the University of Michigan.
Her research focuses on the oldest old and service delivery issues in caregiving.
Tracy Schroepfer is an assistant professor of social work at the Universityof
Wisconsin–Madison. Her research interests include psychosocial needs of terminally
ill elders and their families, interventions to address aggressive behavior displayed
by individuals with Alzheimer’s disease, homeless elders, and caregiver networks.
558 RESEARCH ON AGING
