Full text of DOI:10.1093/geronb/57.1.S14

Journal of Gerontology: SOCIAL SCIENCES
Copyright 2002 by The Gerontological Society of America
2002, Vol. 57B, No. 1, S14–S22
Self-Gain and Self-Loss Among African American
and White Caregivers
Kristie Long Foley,1 Ho-Jui Tung,2 and Elizabeth J. Mutran3
1Department of Public Health Sciences, Wake Forest University School of Medicine, Winston-Salem, North Carolina.
2Department of Health Behavior and Health Education and 3Center on Minority Aging,
University of North Carolina-Chapel Hill.
Objectives. This study compared the effects of a traditional ideology of care and role conflict on the intrinsic rewards
(self-gain) and consequences (self-loss) of caregiving among African Americans and Whites.
Methods. Using a cross-sectional telephone survey of caregivers in North Carolina (n ꢀ 481), the authors performed
a structural equation groups analysis to assess the equality of an a priori caregiving model for African Americans (n ꢀ
257) and Whites (n ꢀ 224).
Results. Despite a stronger preference for family care among African Americans, traditional caregiving ideology
was associated with more self-gain among Whites only; there was no relationship between preference for family care
and self-loss for either group. Furthermore, role conflict was unrelated to self-gain but was related to more self-loss for
both groups. Three additional relationships differed between African Americans and Whites: age and self-gain, gender
and self-gain, and care recipient depression and role conflict. However, 12 proposed relationships were statistically sig-
nificant and equivalent for African Americans and Whites.
Discussion. The findings suggest that contextual elements that influence preference for family care and role conflict
are almost identical for African Americans and Whites. Caregiver demographics are associated with caregiving ideol-
ogy, whereas care recipient frailty is related to role conflict. However, there are differences between African American
and White caregivers, which primarily stem from the role of age, gender, and preference for family care with self-gain.
(Stuckey et al., 1996; Stull et al., 1994). Stuckey and col-
leagues (1996) found that gender of the caregiver and be-
havioral symptoms of the care recipient (CR) were almost
equally related to burden and subjective well-being. In a
study of husband caregivers, memory and behavioral prob-
lems of the wives and use of emotion-focused coping by the
caregiver were solely related to burden (Kramer, 1997a).
However, lower education and problem-focused coping
were related to more satisfaction. Satisfaction in social par-
ticipation was predictive of both positive and negative as-
pects of caregiving.
In another study, positive affect and depression were as-
sessed simultaneously among relative caregivers (Lawton et
al., 1991). Caregiver satisfaction and better health were as-
sociated with positive affect for spouses, whereas poor
health and greater burden were related to their depression.
In contrast, adult children experienced more depression as
they increased their level of help. The authors concluded
that different factors were associated with positive affect
and depression, which may be influenced by the kin rela-
tionship of the caregiving dyad.
AREGIVERS of frail elders are vulnerable to psycho-
logical distress such as depression, stress, increased
C
role strain, loss of identity, and burden (Baillie, Norbeck, &
Barnes, 1988; Burns, Archbold, Stewart, & Shelton, 1993;
Lawton, Moss, Kleban, Glicksman, & Rovine, 1991; Mc-
Fall & Miller, 1992; Skaff & Pearlin, 1992; Thompson, Fut-
terman, Gallagher-Thompson, Rose, & Lovett, 1993; see
Young & Kahana, 1995, for an exception). Although a pes-
simistic perspective has dominated the caregiving literature,
providing care to a loved one may also provide intrinsic re-
wards, closer kinship ties, and self-efficacy in mastering
caregiving tasks (Kramer, 1997b). As Seltzer and Green-
berg (1999, p. 363) contended, “In fact, for many care-
givers, there are positive consequences of assuming this
role, such as feelings of gratification that emanate from
helping a loved one or a renewed sense of purpose in life.
Such positive consequences exist side by side with the more
negative outcomes that predominate.”
Although positive and negative aspects of caregiving
may be aligned, they appear to tap unique components of
the stress and coping process (Stuckey, Neundorfer, &
Smyth, 1996; Stull, Kosloski, & Kercher, 1994). Studies
specifically examining well-being and burden and their cor-
relates have shown that these constructs are correlated.
However, there is some evidence that factors associated
with caregiver gain and strain are distinct.
Ethnicity is another important contextual realm that
shapes the caregiving experience. Evidence suggests that
African Americans are at less risk than Latinos or Whites of
negative caregiving outcomes (Aranda & Knight, 1997;
Fredman, Daly, & Lazur, 1995; Haley et al., 1995, 1996;
Hinrichsen & Ramirez, 1992; Horwitz & Reinhard, 1995;
Lawton, Rajagopal, Brody, & Kleban, 1992). This finding
persists despite the fact that African Americans care for
Caring for someone with greater functional, behavioral,
or cognitive impairment was more strongly correlated to the
negative versus positive aspects of caregiving in two studies
S14

FAMILY CARE, ROLE CONFLICT, SELF-LOSS, AND SELF-GAIN
S15
equally or more frail elderly persons than Latinos and
Whites do (Fredman et al., 1995; Horwitz & Reinhard,
1995; Lawton et al., 1992).
Taking care of one’s own may also reflect a generational
phenomenon. In an ethnographic examination of multigen-
eration African American families, Burton and Sörensen
(1993, p. 53) compared caregivers of different ages. A 64-
year-old caregiver rebuffed the interviewer’s questions
about caregiving stating, “My kin, my friends, all the people
my age know this is what they should do. . . . You never
question it, you just do it.” This is contrasted with a 40-
year-old who feels conflicted about her responsibility as a
caregiver. She receives messages from her mother’s genera-
tion that it is dutiful to take care of one’s own, as well as
messages from her son’s generation that she is merely re-
sponsible for herself.
One explanation that has been offered for this disparity is
that African American caregivers are protected from nega-
tive mental health consequences because of their extended
kin and social networks (Fredman et al., 1995; Horwitz &
Reinhard, 1995). They have and utilize a large number of
informal supports that extend beyond the conjugal family,
such as second- and third-degree relatives, neighbors, and
church parishioners (Wood & Wan, 1993). Flexible kinship
boundaries afford African Americans with greater caregiv-
ing resources and, thus, diffuse the burden.
However, the perception that African Americans are pro-
tected from negative consequences of caregiving because of
extended kinship networks and social support has not been
substantiated (Wood & Wan, 1993). Among adult African
American noncaregivers, depressive symptoms remained
after frequency of contact with network members, durability
of network relationships, and reciprocity of network rela-
tionships were controlled (Thomas, Milburn, Brown, & Gary,
1988). Among caregivers, Haley and colleagues (1995)
found that Whites reported more depression and less self-
efficacy in caring for family members with Alzheimer’s dis-
ease compared with African Americans but that African
Americans did not have more social supports, satisfaction of
support, or more visits with friends and relatives than
Whites. Additionally, social network size and other mea-
sures of social support only marginally reduced the differ-
ence between African Americans’ and Whites’ burden in
caring for adults with mental illness (Stueve, Vine, &
Struening, 1997).
Others have argued that it is not the presence of kinship
ties but rather different appraisal and coping styles that lead
to unique valuations of the caregiving role (Goode, Haley,
Roth, & Ford, 1998; Haley et al., 1996; Horwitz & Rein-
hard, 1995; Knight, Silverstein, McCallum, & Fox, 2000).
For example, caregiver appraisals and coping responses me-
diated the relationship between caregiver race and well-
being (Haley et al., 1996). Whites were more likely than Af-
rican Americans to appraise self-care and behavioral and
memory deficits as more stressful and to use more diverse
coping strategies. Additionally, Goode and colleagues (1998)
found that approach coping (such as positive reappraisal
and seeking guidance and support) was related to better
mental and physical health over time. However, descriptive
statistics showed no difference in coping strategies by race.
Another explanation that has been put forth for racial dif-
ferences is the caregiving ideology of “taking care of one’s
own” (Wood & Wan, 1993; Wright & Mindel, 1993). This
notion of duty in the African American community has been
one of the most relied upon, yet untested, hypotheses. Exis-
tentialists would argue that persons have a choice to assume
the caregiver role but that the choice to take care of one’s
own is shaped by moral responsibility (Farran, 1997). How-
ever, this moral obligation has been discounted by some as a
cultural myth, which may be used as an excuse by the for-
mal health care delivery system to avoid provision of much-
needed health care services to ethnic families (Wright &
Mindel, 1993).
Lawton and colleagues (1992) tested the taking care of
one’s own hypothesis in African American and White care-
givers. African American caregivers reported a more favor-
able caregiving experience than Whites across several indi-
cators: caregiving mastery, traditional caregiving ideology
(the take care of one’s own philosophy), satisfaction, sub-
jective burden, and intrusion on life. However, traditional
ideology was unrelated to depression or positive affect for
either group and was positively related to caregiving satis-
faction among Whites only. However, most studies have ig-
nored the role of family values or ideology in explorations
of caregiver outcomes.
Theoretical Framework
The evolution of caregiving literature on appraisal, cop-
ing, and mental health has led to a reevaluation of the stress
coping paradigm (Knight et al., 2000; Kramer, 1997a). This
reevaluation builds upon the mainstays of stress and coping
frameworks: (a) context in which care occurs (e.g., age,
gender, and socioeconomic status of the caregiver); (b) de-
mands or stressors of the caregiving situation (e.g., behav-
ioral problems, cognitive decline, or physical frailty of the
CR); (c) the caregiver’s appraisal of the situation (e.g., satis-
faction or burden); (d) mediating factors (e.g., social sup-
port and coping); and (e) outcomes of the caregiver, CR, or
both (e.g., mental health, physical health, and long-term
care placement; Lazarus & Folkman, 1984; Pearlin, Mullan,
Semple, & Skaff, 1990).
In 1997, Kramer (1997b) proposed a three-domain adap-
tation model comprising the background and context of
caregiving; the intervening processes, including resources
and appraisal of the caregiving role; and well-being out-
comes. This model was developed from a review of the lit-
erature on positive aspects of caregiving and addressed a
gap in the theoretical understanding of caregiver gain
(Kramer, 1997b; Miller & Lawton, 1997). Kramer proposed
that positive indicators of well-being are especially relevant
to gerontological researchers and that greater attention
should be given to the intervening processes of role gains
and strains on caregiving outcomes.
Furthermore, researchers have extended the stress-coping
paradigm to incorporate a sociocultural dimension (Aranda
& Knight, 1997; Knight et al., 2000). Knight and colleagues
(2000) argued that different ethnic groups, as a result of mi-
nority group membership, are exposed to unique caregiving
experiences. Structural challenges imposed by society (e.g.,
access to formal sector services) and cultural identity (e.g.,

S16
LONG FOLEY ET AL.
familial obligations) influence how ethnicity interacts with
other status variables, the demands and appraisals of care-
giving, and coping style to influence physical and mental
health outcomes. Pearlin and Zarit (1993, p. 157) argued
that comparisons of caregiving outcomes among ethnic
groups are “one of the best ways to discern the social and
cultural circumstances that undergird patterns of care-
giving.”
We used the work by Kramer (1997b) and Knight and
colleagues (2000) within a structural equation modeling
(SEM) framework to simultaneously investigate self-gain
and self-loss for African Americans and Whites. Our pur-
pose in this study was to evaluate two intervening measures
of caregiving appraisal on positive and negative outcomes.
Appraisal was measured by preference for family care,
which represents a traditional caregiving ideology, and role
conflict, which measures the perception that caregiving is
discordant with work and family obligations. The context of
caregiving was framed by characteristics of the care pro-
vider (age, education, income, gender, and role incum-
bency) and CR (deficits in activities of daily living [ADLs]
and instrumental activities of daily living [IADLs] and pres-
ence of disease).
sample would be heterogeneous by race and geographic
location.
For the community-dwelling sample used in this study,
we drew a simple random sample of people aged 65 and
older from each county using the Health Care Financing
Administration’s Medicare Beneficiary Utilization tape. A
total of 4,236 names were provided, of which 904 (21%)
could not be located. Of the remaining 3,332 persons, 411
(12%) refused to be screened for eligibility. To be eligible,
the elderly person must require assistance in one or more
ADLs or IADLs. Of the 2,921 elderly persons screened,
71% were found to be ineligible for the following reasons:
they were not frail (57%); they lived in an institution, had a
Department of Social Services address, or had moved out of
the study area (10%); or they were deceased (4%). We ran-
domly sampled 683 persons of the remaining 852 elderly
persons; of these, 8 refused to participate and 138 were un-
able to complete the questionnaire because of mental in-
competence (a score less than 4 on the Short Portable Men-
tal Status Questionnaire) or physical frailty (unable to
verbally communicate; Pfeiffer, 1975). Thus, a total of 537
elderly persons completed the telephone interview.
All caregivers of the original sample of 683 elderly per-
sons were eligible to participate in the study regardless of
whether or not the frail elderly person participated. Elderly
persons identified 563 family caregivers, and 507 (90%)
completed an approximately 45-min telephone interview.
The principal investigator of this study trained the graduate
students who completed the interviews. Of the caregivers
called, 6.4% refused to participate and 3.5% could not be
contacted or could not verbally communicate. All of the
caregivers, each of whom cared for only one elder in the
study, provided informed consent over the telephone. No
minimum caregiving criteria were necessary to be included
in the study. Caregivers were eligible to participate if the
older person was frail, the older person identified him or her
as the primary caregiver, and the caregiver agreed to partic-
ipate in the study. Prior to analysis, the following persons
were excluded from the study population: 4 Native Ameri-
can caregivers, 1 Asian caregiver, 2 Latin American care-
givers, and 8 who did not provide ethnicity. Additionally,
11 cases were excluded because of excessive missing data.
Five were excluded because they missed one or both of the
self-loss variables, 2 because they missed three or more of
the self-gain variables, and 4 because they missed three or
more of the family care preference variables. The final sam-
ple consisted of 257 African American and 224 White care-
givers (n ꢀ 481).
On the basis of the literature reviewed and the theoretical
framework, our hypotheses were as follows: (a) African
American caregivers will have less role conflict and stron-
ger preferences for family care than White caregivers; (b)
role conflict and preference for family care will be influ-
enced by the characteristics of the caregiver across both eth-
nic groups: age, education, income, gender, and role incum-
bency; (c) greater frailty (ADL/IADL deficits) will have a
direct effect on self-loss; (d) caregivers who provide assis-
tance for individuals diagnosed with Alzheimer’s disease
will experience greater self-loss; (e) family care preference
will be especially important for Whites in increasing self-
gain and equally important for both races in reducing self-
loss; and (f) role conflict will have a direct and positive ef-
fect on self-loss for both groups.
METHODS
Sample
This research was done in response to a call for studies
by the National Institute of Nursing Research to compare
the use and non-use of long-term care among ethnically
diverse groups. Frail elderly persons were selected from
counties in North Carolina that met the following criteria:
minority population greater than 10%, population size
greater than 50,000, and at least five family care homes
(FCH) per county. These criteria, which were met in 29
counties, ensured an ethnically diverse sample of commu-
nity-dwelling and long-term care residents (ꢀ6 per FCH).
The counties were stratified into four groups by total pop-
ulation and minority representation (urban high minority,
urban low minority, rural high minority, and rural low mi-
nority). After the counties were stratified, we randomly
selected 20 counties (4 counties from each urban and 6
from each rural stratum) and selected from them two sam-
ples of frail elderly persons, community-dwelling and
long-term care residents. This strategy ensured that our
Family Caregiver Measures
Caregiving context.—The context of care included char-
acteristics of the caregiver: age (continuous measure), gen-
der (1 ꢀ female), education (six categories ranging from
1 ꢀ “grade school” to six ꢀ “postbaccalaureate”), income
(eight categories ranging from 1 ꢀ “ꢁ$7,500” to 8 ꢀ
“ꢂ$100,001”), and role incumbency (a cumulative role
score counting marriage, employment, and child living at
home). CR characteristics included a summary score for
ADLs and IADLs. We created these scales by asking the

FAMILY CARE, ROLE CONFLICT, SELF-LOSS, AND SELF-GAIN
S17
caregiver if CR needed help with ADLs (dressing, feeding,
bathing, performing personal hygiene, transferring, moving
around the house, and toileting) and IADLs (getting to
places out of walking distance, preparing meals, grocery
shopping, doing housework, doing laundry, taking care of
finances, and obtaining and regulating medicine).
Furthermore, we asked the caregiver if CR had ever seen
a doctor for 17 conditions ranging from visual impairment
to epilepsy. Four conditions were included in the analysis
that were potentially emotionally difficult for the caregiver
and were prevalent enough for us to detect statistically sig-
nificant differences between persons with and without the
disease: Alzheimer’s disease, stroke, cancer, and depres-
sion. We included four specific illnesses, rather than a sum-
mary score of illnesses, to gain as much information about
the role of specific diseases on the caregiving context, yet
maximize statistical power (Ferraro & Wilmoth, 2000). Al-
though we recognized that some CRs will not or cannot
seek treatment for a condition, there was no question that
asked about the mere presence of a disease.
lenges with ADLs, problem behaviors, social contact, and
work status of the caregiver were related to self-loss (ꢃ ꢀ
.76) in a study of spouses and adult children caring for a rel-
ative with Alzheimer’s disease (Skaff & Pearlin, 1992).
Skaff and Pearlin (1992) used four questions to measure
the latent construct of self-gain that assessed the personal
growth aspects of caregiving (ꢃ ꢀ .76): “As a result of care-
giving, do you feel you have . . . (a) become aware of your
inner strengths; (b) become more self-confident; (c) grown
as a person; and (d) learned to do things you didn’t do be-
fore?” We also included a fifth item, “How much do you be-
lieve you have learned to deal with a difficult situation?”
Responses ranged from “not at all” (0) to “very much” (3).
The last item originally was one of four questions used to
measure caregiver competence (Skaff & Pearlin, 1992).
However, the very high correlation between this item and
the other four items of self-gain in our sample prompted us
to include it as an indicator of the self-gain construct. Self-
gain has been associated with other measures of self-con-
cept such as self-esteem and optimism but was not cor-
related with self-loss in Skaff and Pearlin’s study of role
engulfment. Both measures of self-loss and self-gain are
specific to the caregiving role and do not represent global
measures of self-concept or identity.
Caregiving appraisal.—Preference for family care was a
latent construct that tapped a caregiver’s perception of fa-
milial responsibility of care versus rest home care (Sudha &
Mutran, 1999). Respondents were given four statements and
asked to respond on a 4-point scale from “disagree a lot” (0)
to “agree a lot” (3). The statements were (a) rest homes are
places where people with no family go; (b) older people are
always better off living with their family and friends than
they are living in a rest home; (c) adult children should have
to take care of their elderly parents; and (d) if a family really
loves a relative they would never place them in a rest home.
This construct was one of three that emerged during a factor
analysis of racial differences in attitudes toward adult care
homes and care by families. Sudha and Mutran (1999)
found that African American caregivers were more likely
than White caregivers to prefer family care. Additionally,
family care preference among caregivers was influenced by
other contextual factors including age, education, stress of
caregiver, and dislike of rest homes.
The model specified the caregiver and CR characteristics
as exogenous variables. Preference for family care and role
conflict were specified as intervening variables. Self-gain
and self-loss were treated as positive and negative and out-
comes of caregiving.
Analysis
We calculated t tests and chi-square tests of association
for continuous and categorical variables to determine differ-
ences among model elements. We used SEM to test our
conceptual model, which allowed us to statistically deter-
mine the differential effects of the variables in our model on
self-loss and self-gain across groups while simultaneously
considering all hypothesized relationships. SEM also al-
lowed us to assess latent constructs of preference for family
care, role conflict, self-loss, and self-gain by including ob-
served indicators. Using observed indicators corrects for the
measurement error in each item, which we could not have
done if we had used a summary scale of the constructs.
All endogenous variables (preference for family care,
role conflict, self-loss, and self-gain) were treated as latent
constructs (Anderson & Gerbing, 1988). The error vari-
ances (disturbances) between self-loss and self-gain were
correlated because we hypothesized that the two constructs
were theoretically related. Although the paths in the dia-
gram represent a “causal” model, they are correlational by
nature of our cross-sectional design. The ordering of the
variables followed the theoretical framework that has been
proposed.
Role conflict.—We measured role conflict by asking the
caregiver to think about how taking care of the CR affected
him or her personally. Four questions were used for the la-
tent construct: (a) not enough time to spend with your fam-
ily; (b) need to take care of CR when you are not feeling
well; (c) conflicts with job demands and taking care of CR;
and (4) problems that family members, such as your spouse
or children, have with CR.
Self-loss and self-gain.—Self-loss and self-gain are psy-
chological constructs that measure the intrinsic rewards and
consequences for self as a result of caregiving (Skaff &
Pearlin, 1992). Self-loss was a latent construct measured by
two items that assessed a sense of identity loss in the care-
giving role. We asked the caregiver the following question:
“As a result of caregiving, would you say that you feel you
are losing . . . (a) a sense of who you are and (b) a part of
yourself?” Response options were “no” (0), “somewhat”
(1), and “yes” (2). Gender, relationship status, age, chal-
We used LISREL 8 to evaluate the different caregiving
experiences by race (Jöreskog & Sörbom, 1993). We calcu-
lated the zero-order correlation among all of the model ele-
ments for preliminary examination of relationships (data
available upon request). We then conducted a group analy-
sis by fitting the model to each subsample, which tested the
moderating effect of race on caregiving outcomes.

S18
LONG FOLEY ET AL.
Table 2. Structural Equation Groups Analysis of Self-Gain and
Self-Loss to Test for Equality in Caregiving for African Americans
First, a cross-group equality constraint was imposed,
which constrained all hypothesized paths in the model, all
error variances of the indicators for the latent constructs,
and the disturbances in the latent constructs to be equal for
African Americans and Whites. The overall model fit was
significant and indicated a poor fit, ꢄ2 (491, n ꢀ 481) ꢀ
878.84.
Second, we performed a series of steps to improve the
model fit. The constraints imposed on the amount of error in
the measurement model and then the disturbances in the
equations were removed. Next, several structural paths were
freed between African Americans and Whites on the basis
of the maximum modification indices (MMI) in the LIS-
REL analysis. The MMI specified the paths that would im-
prove the model fit if allowed to vary across groups (indi-
cating structural differences between the two groups). The
ꢄ2 for each respecified model was compared with the ꢄ2 of
the previous model, and we calculated the differences in ꢄ2
values and degrees of freedom to determine if the modifica-
tion made a significant improvement in model fit.
(n ꢀ 257) and Whites (n ꢀ 224)
Contrast With
Baseline Model
Fit Indices
Model
ꢄ2
df
p
ꢄ2
difference
dfdifference Critical n GFI
Baseline model (all
parameters
constrained to be
equal)
Error terms of
indicators of
endogenous
878.84 491 0.00
—
—
15
5
309.95 0.88
316.35 0.88
330.28 0.90
constructs freed 819.27 476 ꢅ.05 59.57
Allowed covariances
among the latent
constructs to
vary
793.28 471 ꢅ.05 25.99
Path from preference
for family care
→ self-gain freed 776.37 470 ꢅ.05 16.91
Path from gender →
1
1
1
336.79 0.90
339.58 0.90
344.04 0.90
self-gain freed
Path from age →
self-gain freed
Path from depression
→ role conflict
freed
768.46 469 ꢅ.05
7.91
RESULTS
756.97 468 ꢅ.05 11.49
Caregivers
Table 1 reports the descriptive statistics and statistical
differences of the model elements by race. Caregivers were
predominately female and middle-aged with a high school
750.14 467 ꢅ.05
6.83
1
346.47 0.90
Notes: Freeing paths in the structural equation model suggests that these re-
lationships (e.g., family care → self-gain) are different for African American
and White caregivers. GFI ꢀ goodness of fit index.
Table 1. Descriptive Characteristics of the Sample
African
All
(n ꢀ 481)
American
(n ꢀ 257)
White
(n ꢀ 224)
education. African Americans had less income and were
more likely to be single and caring for a child at home. They
also provided more help with ADLs and IADLs, although
White caregivers assisted more persons diagnosed with can-
cer and depression. African American caregivers reported a
higher preference for family versus rest home care, but
fewer roles and less role conflict than White caregivers. Af-
rican Americans were also more likely to report less self-
loss and more self-gain in the caregiving role.
Variable
Age**
M (SD)
Range
Female, %**
Highest degree received*
Less than high
school, %
High school
graduate, %
At least some
college, %
Median income***
Roles
ADLs, M (SD)*
IADLs, M (SD)***
Alzheimer’s
56.47 (14.09)
(15.27–90.14)
75.9
54.87 (14.51)
(16.62–90.14)
81.7
58.30 (13.40)
(15.27–85.08)
69.2
31.19
31.39
37.42
34.63
33.07
32.30
27.23
29.46
43.31
Race and Self-Loss and Self-Gain
The initial step in the SEM analysis compared the con-
strained versus unconstrained models of self-loss and self-
gain. This approach statistically forced the model (and re-
gression coefficients) to be equivalent across racial groups
to test the overall model fit and null hypothesis that African
Americans and Whites share equivalent caregiving experi-
ences. An examination of the chi-square/degrees of freedom
ratio as well as other fit indices suggested that the model
needed to be respecified and that African Americans and
Whites were different (Table 2).
$15,501–25,000 $15,501–25,000 $25,001–35,000
1.43 (0.87)
1.08 (1.93)
3.77 (2.50)
1.37 (0.91)
1.28 (1.99)
4.15 (2.41)
1.51 (0.81)
.85 (1.83)
3.34 (2.53)
disease, %
Stroke, %
8.1
26.8
19.5
15.4
9.7
29.2
13.2
12.1
6.3
24.1
26.8
19.2
Cancer, %***
Depression, %*
Role conflict,
M (SD)**
Preference for family
care, M (SD)***
Self-loss, M (SD)**
Self-gain, M (SD)**
5.87 (1.95)
5.62 (1.73)
6.16 (2.13)
The first respecification allowed the error variances
among the indicators for the latent constructs to vary across
groups. We found a significant improvement in model fit by
freeing these terms. This suggests that the amount of error
variance in the items was not equal across the two racial
groups; however, no pattern emerged that indicated the pro-
posed relationships in the caregiving model were consis-
6.73 (2.69)
0.42 (0.96)
11.58 (3.90)
7.31 (2.70)
.30 (.78)
12.06 (3.72)
6.08 (2.53)
.57 (1.15)
11.03 (4.03)
Notes: ADL ꢀ activities of daily living; IADL ꢀ instrumental activities of
daily living.
*p ꢅ .05; **p ꢅ .01; ***p ꢅ .001.

FAMILY CARE, ROLE CONFLICT, SELF-LOSS, AND SELF-GAIN
S19
was associated with more self-gain. However, the opposite
was found among Whites; older age was related to more
self-gain. Also, White women were more likely than White
men to report self-gain, but there was no gender difference
among African Americans. For all caregivers, having more
roles and caring for someone with cancer were directly as-
sociated with more self-gain. Additionally, more roles and
greater role conflict were associated with self-loss among
all caregivers.
Indirect Effects Via Caregiving Appraisal
Being male and having less education and income were
associated with greater traditional caregiving ideology re-
gardless of race. Additionally, providing care for someone
with greater IADL deficits was associated with a lower
preference for family care. Caregiver gender and IADLs
were not related to self-gain indirectly through family care
preference. However, preference for family care mediated
the relationship between education, income, and self-gain
for Whites only. Those who were less educated and with
lower incomes had a preference for family care regardless
of race, but socioeconomic status only indirectly influenced
self-gain for Whites. The mean differences of income and
education between African Americans and Whites helped to
explain the distinction in family care by race. It also sug-
gested that self-gain among African American caregivers
did not depend on the relationship between socioeconomic
status and caregiving ideology. However, among Whites,
those with lower education and income were more likely to
prefer family care and thus indirectly express more self-gain
through traditional caregiving ideology.
We also found that role conflict mediated the relation-
ships between several contextual factors and self-loss. Be-
ing young and having greater role incumbency were related
to role conflict. Additionally, caring for someone with more
IADL deficits and Alzheimer’s disease was associated with
greater role conflict. White caregivers who provided assis-
tance to an older person previously seen by a health care
provider for depression were more likely to report role con-
flict. However, there was no association with CR depression
and role conflict among African Americans. Role incum-
bency was the only contextual factor that was associated
with self-loss both directly and indirectly through role con-
flict.
Figure 1. Structural model of self-gain and self-loss for African
American (n ꢀ 257) and White (n ꢀ 224) caregivers. Significant un-
standardized regression coefficients (unless otherwise noted with ns
[not significant]) are presented for African Americans (in bold) and
Whites when the equality constraints in the group analysis were re-
laxed. Freeing these structural paths was based on the maximum
modification indices of LISREL and significantly improved the
model fit. The regression coefficients are equal for both groups be-
cause of the imposed equality constraint. All possible paths among
the covariates were tested but are not presented. ADL ꢀ activities of
daily living; IADL ꢀ instrumental activities of daily living.
tently better in one group than the other. Our next step in the
analysis was to free the equality constraints among the dis-
turbances of the endogenous constructs. An improvement in
model fit suggested that the model explained a significantly
different proportion of variance in the equation for African
Americans than for Whites. Thus far, allowing only residual
variances to vary preserved the same structural pattern of
self-loss and self-gain for African Americans and Whites.
The final phase of our analysis was to modify four pa-
rameters in the structural model: (a) preferences for family
care to self-gain, (b) gender to self-gain, (c) age to self-gain,
and (d) depression to role conflict. The final goodness-of-fit
statistics included a chi-square/degrees of freedom ratio of
less than 1.75, a critical n of more than 300, and a goodness-
of-fit index of .90. The unstandardized regression coeffi-
cients are presented in Figure 1 for all statistically signifi-
cant paths and include coefficients for African Americans
(in bold) and Whites where the relationships differed across
racial groups.
DISCUSSION
In this study we examined contributions of traditional
caregiving ideology and role conflict on self-gain and self-
loss among African American and White caregivers. In sup-
port of our first hypothesis, African Americans report sig-
nificantly more traditional caregiving ideology and less role
conflict than Whites before statistical controls are exercised.
Additionally, traditional ideology is positively associated
with self-gain for Whites only and equally unimportant for
both races in reducing self-loss when all hypothesized rela-
tionships are considered.
It has been argued that a moral obligation to assist frail
kin protects African Americans from the mental health con-
sequences of caregiving (Wood & Wan, 1993; Wright &
Mindel, 1993). However, our findings indicate that African
Direct Effects on Caregiving Outcomes
Two sociodemographic factors, age and gender, differed
by racial group. Among African Americans, being younger

S20
LONG FOLEY ET AL.
Americans’ more pervasive traditional values of caregiving
explained neither their lowered sense of self-loss nor the ra-
cial differences in self-loss across groups, a finding that has
some support in the literature (Lawton et al., 1992).
In addition, Whites (especially lower income Whites)
with stronger preferences for family care may experience
self-gain when they adapt their other roles to accommodate
the added responsibilities of caregiving. African Americans
may regard caregiving behavior as central to their lives, re-
gardless of socioeconomic status, and may not give them-
selves extra credit for activities not perceived as a substan-
tial departure from the norm. Thus, the more pervasive
traditional values of care provision neither protect nor bene-
fit African American caregivers.
The factors that influence ideology are the same for Afri-
can Americans and Whites and are primarily related to care-
giver demographics versus care recipient frailty. Men have
a stronger preference for family care, suggesting a gender
difference in caregiving ideology across both groups. Fe-
male caregivers providing assistance to frail men who main-
tain traditional caregiving values may have more guilt if
caregiving values are not shared. Also, caregivers with
more education and income report a lower preference for
family care. However, this does not appear to be related to
the social and economic costs of caregiving given that we
do not find a significant association between education, in-
come, and role conflict.
Similar to traditional ideology, most model elements as-
sociated with role conflict are the same for African Ameri-
cans and Whites. However, these factors primarily reflect
the caregiving demands (e.g., number of roles or caring for
someone with Alzheimer’s disease) versus the demographic
context in which care occurs. Moreover, role conflict is as-
sociated with greater self-loss for both groups and is unre-
lated to self-gain. The only difference between the two
groups is the association between depression and role con-
flict; caring for a relative who has seen a health care pro-
vider for depression is related to greater role conflict for
Whites only. This raises a question regarding how depres-
sion is manifested across the two groups and how depres-
sive symptomatology could lead to different caregiving de-
mands.
The hypotheses that greater ADL/IADL frailty and caring
for someone with Alzheimer’s disease would directly affect
self-loss are unsupported. ADL dependency is unrelated to
any caregiving appraisal or outcome, and IADL impairment
and caring for someone with Alzheimer’s disease are only
indirectly related to self-loss through role conflict. This sug-
gests that the appraisal of the caregiving situation is at least
as important as the tangible demands of caregiving itself.
We find two additional differences between African
American and White caregivers not hypothesized. First,
older African American caregivers report less self-gain,
whereas older White caregivers report more self-gain. These
findings may reflect a frailty among older African Ameri-
can caregivers that leads to less gainful caregiving or gener-
ational differences in caregiving expectations by ethnicity
that influence the opportunity for intrinsic rewards across
the life course.
more self-gain than White men, but there is no gender dif-
ference in self-gain among African Americans. Several
studies have documented gender differences in negative
mental health outcomes among caregivers (Kramer & Kip-
nis, 1995; Miller & Cafasso, 1992; Zarit, Todd, & Zarit,
1986). Additionally, Callahan and Wolinsky (1994) found
differences in depressive symptomatology among chroni-
cally ill racial and gender subgroups, with White women re-
porting the highest levels of depressive symptoms. Our
findings, along with published literature on gender and eth-
nic differences in mental health, raise questions regard-
ing gender-role socialization of mental health. Are White
women exclusively socialized to reap (or admit) the intrin-
sic consequences and rewards of care provision? Does gen-
der-role socialization vary by race? Gender-role socializa-
tion in caregiving deserves further consideration, especially
embedded within other contextual realms of race and socio-
economic status.
This study adds to the discourse on ethnic and cultural dif-
ferences in caregiving appraisals and outcomes across racial
groups. Although African American caregivers report less
self-loss and more self-gain than Whites, traditional ideol-
ogy and role conflict are unrelated to “better” mental health
outcomes for African Americans when all relationships are
tested simultaneously. To our knowledge, this is only the
second study to test the commonly held view that traditional
caregiving values protect African American caregivers from
the psychological consequences of caregiving.
We echo previous studies that have reported the related
yet distinct concepts of loss and gain among caregivers
(Lawton et al., 1991; Stuckey et al., 1996; Stull et al., 1994).
Not only does the context of care uniquely predict the posi-
tive and negative aspects of caregiving, but also caregiver
appraisal differentially affects self-loss and self-gain. In-
cluding positive and negative aspects of caregiving and two
different types of appraisal (one grounded in culture that ex-
ists outside of the caregiving context and one that is situa-
tional and rooted in the caregiving situation) in one study is
an important contribution to the caregiving literature. Al-
though we cannot rule out the possibility that discrepancies
in mental health outcomes found in our study are merely
measurement biases (Dilworth-Anderson & Anderson, 1994),
we tested “more complex models based on theory and con-
textual frameworks that reflect racial, ethnic, and cultural
differences” (Connell & Gibson, 1997, p. 363). Another
strength of this study is the generalizability of the findings.
We included a heterogeneous sample of elderly persons and
their caregivers and oversampled African American care-
givers, addressing some of the major methodological con-
cerns in caregiving research (Baumgarten, 1989).
Although this study adds to the literature on racial differ-
ences in caregiving outcomes, several limitations must be
addressed. First, cultural and ethnic hypotheses about care-
giving experiences could have been more thoroughly evalu-
ated given the richness of our sample. This is probably the
greatest limitation of our study and others that have at-
tempted to explain racial and ethnic differences in caregiver
outcomes. In future studies of this nature, investigators
should ask questions about cultural and ethnic identity as
well as racial prejudice and discrimination. Additionally, in-
Second, we find that White women report significantly

FAMILY CARE, ROLE CONFLICT, SELF-LOSS, AND SELF-GAIN
S21
Fredman, L., Daly, M. P., & Lazur, A. M. (1995). Burden among White
and Black caregivers to elderly adults. Journal of Gerontology: Social
Sciences, 50B, S110–S118.
Goode, K. T., Haley, W. E., Roth, D. L., & Ford, G. R. (1998). Predicting
longitudinal change in caregiver physical and mental health: A stress
process model. Health Psychology, 17, 190–198.
Haley, W. E., Roth, D. L., Coleton, M. I., Ford, G. R., West, C. A. C., Col-
lins, R. P, & Isobe, L. T. (1996). Appraisal, coping, and social support
as mediators of well-being in Black and White family caregivers of pa-
tients with Alzheimer’s disease. Journal of Consulting and Clinical
Psychology, 64, 121–129.
cluding caregiving perspectives on ideology and role con-
flict from different family members could enhance future
caregiving research. Discordant family ideology could af-
fect the mental health of the caregiver and CR. A second
limitation is the limited causal interpretations that can be
made of the findings because of the cross-sectional data in
this study. Future research would be greatly improved and
the findings substantiated by use of longitudinal study de-
signs.
Haley, W. E., West, C. A. C., Wadley, V. G., Ford, G. R., White, F. A.,
Barrett, J. J., Harrell, L. E., & Roth, D. L. (1995). Psychosocial, social
and health impact of caregiving: A comparison of Black and White de-
mentia family caregivers and noncaregivers. Psychology and Aging,
10, 540–552.
Hinrichsen, G. A., & Ramirez, M. (1992). Black and White dementia care-
givers: A comparison of their adaptation, adjustment, and service
utilization. The Gerontologist, 32, 375–381.
Horwitz, A., & Reinhard, S. C. (1995). Ethnic differences in caregiving du-
ties and burden among parents and siblings of persons with severe
mental illnesses. Journal of Health and Social Behavior, 36, 138–150.
Jöreskog, K. G., & Sörbom, D. (1993). LISREL 8. User’s reference guide.
Mooresville, IN: Scientific Software.
Knight, B. B., Silverstein, M., McCallum, T. J., & Fox, L. S. (2000). A so-
ciocultural stress and coping model for mental health outcomes among
African American caregivers in southern California. Journal of Geron-
tology: Psychological Sciences, 55B, P142–P150.
To conclude, the context of care must extend beyond the
tangible role commitments and conflicts of the caregiver if
the sociocultural dimensions of caregiving are to be ad-
dressed (Aranda & Knight, 1997; Knight et al., 2000). We
contribute to the literature on ethnic differences in caregiv-
ing outcomes by examining both self-loss and self-gain
within the context of role conflict and traditional caregiving
ideology for African American and White caregivers. We
find that one measure of taking care of one’s own could
not explain better mental health outcomes among African
Americans, a commonly used but rarely tested ideological
explanation for racial differences in caregiving outcomes.
Acknowledgments
Kramer, B. J. (1997a). Differential predictors of strain and gain among
husbands caring for wives with dementia. The Gerontologist, 37, 239–
249.
A National Institute on Nursing Research grant (RO1 NR 03406; E. J.
Mutran, principal investigator) supported this research.
Kramer, B. J. (1997b). Gain in the caregiving experience: Where are we?
What next? The Gerontologist, 37, 218–232.
Kramer, B. J., & Kipnis, S. (1995). Eldercare and work-role conflict: To-
ward an understanding of gender differences in caregiver burden. The
Gerontologist, 35, 340–348.
Address correspondence to Kristie M. Long Foley, Wake Forest Uni-
versity School of Medicine, Department of Public Health Sciences,
Piedmont Plaza II, Suite 512, Winston-Salem, NC 27157. E-mail: kfoley
@wfubmc.edu
Lawton, M. P., Moss, M., Kleban, M. H., Glicksman, A., & Rovine, M.
(1991). A two-factor model of caregiving appraisal and psychological
well-being. Journal of Gerontology: Psychological Sciences, 46,
P181–P189.
Lawton, M. P., Rajagopal, D., Brody, E., & Kleban, M. H. (1992). The dy-
namics of caregiving for a demented elder among Black and White
families. Journal of Gerontology: Social Sciences, 47, S156–S164.
Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. New
York: Springer.
McFall, S., & Miller, B. H. (1992). Caregiver burden and nursing home ad-
mission of frail elderly persons. Journal of Gerontology: Social
Sciences, 47, S73–S79.
Miller, B., & Cafasso, L. (1992). Gender differences in caregiving: Fact or
artifact? The Gerontologist, 32, 498–507.
References
Anderson, J. C., & Gerbing, D. N. (1988). Structural equation modeling in
practice: A review and recommended two-step approach. Psychologi-
cal Bulletin, 103, 411–423.
Aranda, M. P., & Knight, B. G. (1997). The influence of ethnicity and cul-
ture on caregiver stress and coping process: A sociocultural review and
analysis. The Gerontologist, 37, 342–354.
Baillie, V., Norbeck, J. S., & Barnes, L. E. A. (1988). Stress, social sup-
port, and psychological distress of family caregivers of the elderly.
Nursing Research, 37, 217–222.
Baumgarten, M. (1989). The health of persons giving care to the demented
elderly: A critical review of the literature. Journal of Clinical Epidemi-
ology, 42, 1137–1148.
Miller, B., & Lawton, M. P. (1997). Positive aspects of caregiving. Intro-
duction: Finding balance in caregiver research. The Geronotologist, 37,
216–217.
Pearlin, L. I., Mullan, J. T., Semple, S. J., & Skaff, M. M. (1990). Caregiv-
ing and the stress process: An overview of concepts and their measures.
The Gerontologist, 30, 583–594.
Pearlin, L. I., & Zarit, S. H. (1993). Research into informal caregiving:
Current perspectives and future direction. In S. H. Zarit, L. I. Pearlin, &
K. W. Schaie (Eds.), Caregiving systems: Informal and formal helpers
(pp. 155–167). Hillsdale, NJ: Erlbaum.
Burns, C., Archbold, P., Stewart, B., & Shelton, K. (1993). New diagnosis:
Caregiver role strain. Nursing Diagnosis, 4, 70–74.
Burton, L. M., & Sörensen, S. (1993). Temporal context and the caregiver
role: Perspectives from ethnographic studies of multigeneration Afri-
can-American families. In S. H. Zarit, L. I. Pearlin, & K. W. Schaie
(Eds.), Caregiving systems: Informal and formal helpers (pp. 47–66).
Hillsdale, NJ: Erlbaum.
Callahan, C. M., & Wolinsky, F. D. (1994). The effects of gender and race
on the measurement properties of the CES-D in older adults. Medical
Care, 32, 341–356.
Pfeiffer, E. (1975). A short portable mental status questionnaire for the as-
sessment of organic brain defects in elderly patients. Journal of the
American Geriatics Society, 23, 433–441.
Connell, C. M., & Gibson, G. D. (1997). Racial, ethnic, and cultural differ-
ences in dementia caregiving: Review and analysis. The Gerontologist,
37, 355–364.
Seltzer, M. M., & Greenberg, J. S. (1999). The caregiving context: The in-
tersection of social and individual influences in the experience of
family caregiving. In C. D. Ryff & V. W. Marshall (Eds.), The self and
society in aging processes (pp. 362–397). New York: Springer.
Skaff, M. M., & Pearlin, L. I. (1992). Caregiving: Role engulfment and the
loss of self. The Gerontologist, 32, 656–664.
Stuckey, J. C., Neundorfer, M. M., & Smyth, K. A. (1996). Burden and
well-being: The same coin or related currency? The Gerontologist, 36,
685–693.
Dilworth-Anderson, P., & Anderson, N. B. (1994). Dementia caregiving in
Blacks: A contextual approach to research. In E. Light, N. Niederehe,
& B. Lebowitz (Eds.), Stress effects on family Alzheimer’s patients:
Research and interventions (pp. 385–409). New York: Springer.
Farran, C. J. (1997). Theoretical perspectives concerning positive aspects
of caring for elderly persons with dementia: Stress/adaptation and exis-
tentialism. The Gerontologist, 37, 250–256.
Ferraro, K. F., & Wilmoth J. M. (2000). Measuring morbidity: Disease
counts, binary variables, and statistical power. Journal of Gerontology:
Social Sciences, 55B, S173–S189.
Stueve, A., Vine, P., & Struening, E. L. (1997). Perceived burden among

S22
LONG FOLEY ET AL.
caregivers of adults with serious mental illness: Comparison of Black,
Hispanic, and White families. American Journal of Orthopsychiatry,
(Eds.), Ethnic elderly and long-term care (pp. 39–56). New York:
Springer.
67, 199–209.
Wright, R., Jr., & Mindel, C. H. (1993). Economic, health and service use
policies: Implications for long-term care of ethnic elderly. In C. M.
Barresi & D. E. Stull (Eds.), Ethnic elderly and long-term care (pp.
247–263). New York: Springer.
Stull, D. E., Kosloski, K., & Kercher, K. (1994). Caregiver burden and ge-
neric well-being: Opposite sides of the same coin. The Gerontologist,
34, 88–94.
Sudha, S., & Mutran, E. J. (1999). Ethnicity and eldercare: Comparison of
attitudes toward adult care homes and care by families. Research on
Aging, 21, 570–594.
Young, R. F., & Kahana, E. (1995). The context of caregiving and well-
being outcomes among African and Caucasian Americans. The Geron-
tologist, 35, 225–232.
Thomas, V. G., Milburn, N. G., Brown, D. R., & Gary, L. E. (1988). Social
support and depressive symptoms among Blacks. The Journal of Black
Psychology, 14, 35–45.
Zarit, S. H., Todd, P., & Zarit, J. (1986). Subjective burden of husbands
and wives as caregivers: A longitudinal study. The Gerontologist, 26,
260–266.
Thompson, E. H., Jr., Futterman, A. M., Gallagher-Thompson, D., Rose, J.
M., & Lovett, S. B. (1993). Social support and caregiving burden in
family caregivers of frail elders. Journal of Gerontology: Social Sci-
ences, 48, S245–S254.
Wood, J. B., & Wan, T. T. H. (1993). Ethnicity and minority issues in fam-
ily caregiving to rural Black elders. In C. M. Barresi & D. E. Stull
Received November 21, 2000
Accepted May 21, 2001
Decision Editor: Fredric D. Wolinsky, PhD