Professional Documents
Culture Documents
Untitled
Untitled
Well-being
Dissertation
Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy
By
Seojin Won
2009
Dissertation Committee:
Mo-Yee Lee
Holly Dabelko-Schoeny
Copyright by
Seojin Won
2009
Abstract
alternative child care support and serve as role models for grandchildren. Although social
workers have increasingly examined the needs of custodial grandparents who face many
who are important supports for families; this role of grandparents is especially crucial as
rates rise, mothers increasingly seek alternative caregivers for their children. They often
choose grandparents as alternative caregivers, especially when children are young. While
both custodial and noncustodial grandparents struggle with the burden of childcare, the
Thus, the purpose of this study was to identify factors associated with the
closeness between the grandparent and the grandchild. Also, the researcher sought factors
grandparent’s well-being. This study conducted secondary data analysis using the third
wave of the National Survey of Families and Households (NSFH) dataset. The number of
grandparents in the dataset was 2,541. Structural Equation Modeling (SEM) was applied
ii
Findings indicated that older grandparents tended to interact less with their
with their grandchild without the parent of the child, the well-being of the grandparents
meeting, and the grandparent’s self-rating of closeness with his or her grandchild were
interacted with their grandchild more frequently had a higher level of well-being than
those who had less contact with their grandchild. Thus, it is assumed that the
burden for them. In this study, spending nights with the grandchild without the parent of
the child seems to be the point where the grandparents start feeling burdened. This
finding was consistent regardless of grandparents’ gender, ethnicity, and the presence or
absence of legal custody. This study suggested that social support should target both
great effort into child care feel the same kind of stress as custodial grandparents. In
because various factors are associated with the well-being of the grandparents.
iii
Dedication
Dedicated to my parents
iv
Acknowledgements
Many faculty members, friends, and family have supported me in completing this
dissertation. Without their support and encouragement, I never would have finished this
dissertation process.
experience helped me to think more critically and to become a better scholar. I would
especially like to express sincere gratitude to my advisor, Dr. Virginia Richardson, for
her endless encouragement and consistent support of this dissertation. She has always
been my biggest supporter throughout my academic career at The Ohio State University.
She spent many hours reading drafts of this dissertation and proposal for feedback. I also
appreciate all the support, encouragement, and advice of Dr. Mo-Yee Lee. She helped me
establish a more thorough theoretical framework for this dissertation. I also would like to
thank Dr. Holly Dabelko-Schoeny for her continued support and feedback throughout the
process.
I wish to thank Jennifer Nakayama for her help and support. She helped me to
successfully process all the administrative work since I entered into the doctoral program.
And, she was an invaluable resource from whom I could get answers for issues relating to
v
I am grateful to my colleagues, Wonik Lee, Se Kyung Moon, Jinhyun Kim, and
Jina Han, who provided advice and feedback on statistical analyses whenever I asked.
valuable input made this dissertation more comprehensive. In addition, their emotional
I want to thank my friend, Ohyon Paek, for everything she has done for me so far.
She has been my best friend, my sister, and my supporter. She has stimulated me to keep
Lastly, I am grateful to my mother and father for their endless love and support. I
cannot express how thankful I am. I would never be able to continue my studies at the
master’s and doctoral levels without their trust and dedication. They have guided me to
be a good researcher.
vi
Vita
2002 …………..….. B.A. Social Welfare, Seoul National University, Republic of Korea
Publications
Fields of Study
vii
Table of Contents
Abstract ............................................................................................................................... ii
Dedication .......................................................................................................................... iv
Acknowledgements..............................................................................................................v
Vita.................................................................................................................................... vii
Definition of Grandparents...........................................................................................5
Consequences of Grandparenting...............................................................................15
Rationale.....................................................................................................................42
Research Questions....................................................................................................44
viii
Chapter 4: Research Methods ............................................................................................46
Dataset .......................................................................................................................46
Sample .......................................................................................................................49
Initial Analyses...........................................................................................................62
References .......................................................................................................................114
ix
List of Tables
Table 1.1. Child Care Arrangement among Working Mothers with Children under 5
Table 1.2. Labor Force Participation among Mothers with Children ..................................2
Table 5.5. Grandparent’s Functional Capacity: Physical and/or Mental Limits ...............67
Table B.5.1. Covariance Matrix Used for Hypothesis Test 5 (Caucasian American
Grandparents) ...........................................................................................................137
Table B.5.2. Covariance Matrix Used for Hypothesis Test 5 (Non-Caucasian American
Grandparents) ...........................................................................................................137
Table B.6.1. Covariance Matrix Used for Hypothesis Test 6 (Custodial Grandparents).138
Table B.6.2. Covariance Matrix Used for Hypothesis Test 6 (Noncustodial Grandparents)
..................................................................................................................................138
xi
List of Figures
Youngest Grandchild, and Age of the Oldest Grandchild on the Nights Spent with
Grandchild and the Closeness between the Grandparent and the Grandchild............75
Figure 5.2. Impact of Nights Spent with Grandchild and the Closeness between the
Functional Capacity on the Relationship of the Nights Spent with Grandchild, the
Closeness between the Grandparent and the Grandchild, and the Grandparent’s Well-
being ...........................................................................................................................81
Functional Capacity on the Relationship of the Nights Spent with Grandchild, the
Closeness between the Grandparent and the Grandchild, and the Grandparent’s Well-
Functional Capacity on the Relationship of the Nights Spent with Grandchild, the
Closeness between the Grandparent and the Grandchild, and the Grandparent’s Well-
xii
Figure 5.6. Effects of Cultural Attitude, Grandparent-parent Support, and Grandparent’s
Functional Capacity on the Relationship of the Nights Spent with Grandchild, the
Closeness between the Grandparent and the Grandchild, and the Grandparent’s Well-
Functional Capacity on the Relationship of the Nights Spent with Grandchild, the
Closeness between the Grandparent and the Grandchild, and the Grandparent’s Well-
Functional Capacity on the Relationship of the Nights Spent with Grandchild, the
Closeness between the Grandparent and the Grandchild, and the Grandparent’s Well-
Functional Capacity on the Relationship of the Nights Spent with Grandchild, the
Closeness between the Grandparent and the Grandchild, and the Grandparent’s Well-
Figure 6.1. Key Concepts of Life Course Perspective and Variables in This Study ......102
Figure 6.2. Key Concepts of Bowen Family Systems Theory and Variables in This Study.
..................................................................................................................................104
xiii
Chapter 1: Introduction
The growing senior population and increasing maternal employment rate in the
U.S. has resulted in more senior citizens becoming caregivers for their grandchildren.
The proportion of seniors in the U.S. population had gradually increased from 6.8% in
1940 to 11.3% in 1980, and to 12.4% in 2000 (U.S. Census Bureau, 2006). The recent
annual population estimates indicated that the proportion of senior population remained
almost same compared to the proportion in 2000 as it was 12.53% in 2007 (U.S. Census
Bureau, 2007). As the senior population grows, many older adults are becoming
reported that they were caregivers of their grandchildren while the parents were at work.
The AARP also identified that 24% of those who take care of their grandchildren while
the parents are at work also provide childcare when the parents are not at work. Similarly,
Table 1.1. shows that 28.3% of mothers chose children’s grandparents as caregivers in
2002, followed by the father (24.7%), multiple arrangements (22.4%), or day care centers
(20.6%). When children are young, more mothers tend to choose grandparents as
1
Grand- Day care Sibling / Family Multiple
Father
parent center other relative daycare arrangement
Besides the fact of senior population growth, the increase in labor force
participation among mothers may affect the tendency of older adults to become childcare
providers. According to the U.S. Census Bureau (2009), 61.8% of mothers who were 16
years old and over were in the workforce in 2007. More specifically, Table 1.2. shows
As described in Table 1.2., 32.7% of mothers with children under three years old
were in the workforce in 1975, and the percentage peaked to 59% in 2000. After 2000,
the labor force participation rate among mothers with children under 3 years old had
2
remained almost the same. Labor force participation among mothers with older children
was much higher. This tendency may affect the living arrangements of older adults as
well. According to the U.S. Census Bureau (2003), 3.6% of people who were 30 years
old or older lived with their grandchildren. Among those grandparents living with their
grandchildren, 2.4 million (42%) grandparents were responsible for meeting the basic
needs of their grandchildren. Moreover, 39% of them had cared for their grandchildren
for more than 5 years. The results also indicated that 62.7% of grandparents who were
responsible for their grandchildren were female. In regards to the age of grandparents,
35.1% of custodial grandparents were in their 50’s, 29.2% were in their 40’s and 21.0%
were in their 60’s. Based on the census results, the majority of grandparents serving as
the primary caregivers for their grandchildren were grandmothers between 40 and 69
years old. The American Association of Retired People (AARP) (2006b) also stated,
based on the 2006 census results, that more than 4.5 million children might currently
reside in households headed by grandparents. This means that one out of twelve children
Presser (1989) stated that grandmothers and mothers in the study tended to
grandmothers worked completely different hours from the child’s mother in order to care
for the child. In addition, one-fifth of grandmothers had only a little overlap of working
hours with the mother’s working hours. Wheelock and Jones (2002) stated that full-time
working parents in the United Kingdom tended to utilize both formal and informal child
care. For instance, 22% of parents who used regular informal child care also used regular
3
formal child care. In addition, 18% of parents who used regular formal child care also
As grandparents support their adult child by serving as a third party in the family
system, their well-being may be influenced by their relationship with the members of
their adult child’s family. Although grandparent caregivers often report physical, mental,
through the grandparent-grandchild relationship (Hayslip Jr. & Kaminski, 2005). In this
study, the researcher focused on the closeness between grandparents and grandchildren
and its effect on the grandparents’ well-being. In addition, differences between male
Likewise, the differences among ethnic groups and the differences between custodial
addressed.
4
Chapter 2: Literature Review
Definition of Grandparents
they assume in caring for their grandchild. First of all, ‘custodial grandparents’ are those
who have legal custody of their grandchildren. In other words, they are responsible for
daily life and the decision-making process. In addition, they are able to receive child care
custodial grandparents, in that they may take care of their grandchild, but they do not
have legal custody (Pruchno & McKenney, 2000). Hence, non-custodial grandparents
Targ (1998) stated that ‘living-with-grandparents’ are those who provide daily care for
their grandchildren but do not have legal custody. Approximately 70% of the
grandparents in their study lived with their grandchild and the parent of the child.
‘Daycare grandparents’ are those who help the child’s parents to fulfill their own needs.
This population is close to the societal definition of grandparent since the children return
to their parents at the end of the day. Traditional grandparents are similar to the daycare
grandparents whose grandchildren are in the care and custody of their own parents, yet
the grandparents sometimes help the parents care for the children.
5
Factors Affecting the Grandparent-grandchild Relationship
The family is increasingly considered to be a social system that is affected by, and
that affects, both internal and external factors (Ashford, LeCroy, & Lortie, 2001).
Grandparents are often assumed to be either formal or informal childcare providers for
grandchild include gender, the developmental stage of the grandchild, the relationship
between the grandparent and the parent, and the relationship between the parent and the
child. External factors may include the culture of the ethnic group and social support for
grandparenting.
Gender
Women are often perceived as the kin-keepers who maintain relationships with other
family members (Lemme, 2002). Hence, grandmothers tend to have more positive
grandparent identities than grandfathers (Reitzes & Mutran, 2004). Reitzes and Mutran
assumed that the reason for women’s more positive identities is the traditional gender
norms for women to be caregivers. Strom and Strom (1997) identified perceptions that
Hispanic American families had of each other. In this study, they discovered that
Caucasian American grandmothers were seen as more satisfied, more successful, and
6
Maccoby (2003) addressed the effect of gender on interactions between the parent
and the child. Maccoby indicated that parents tended to spend more time and have a more
intimate relationship with the same-sex child. Similarly, in a study by Crouter, McHale,
and Bartko (1993), mothers were equally involved with daughters and sons, while fathers
were more involved with their sons. Although it is not clear if the differences in
interaction, the power structure within the family may support Maccoby’s findings. In
other words, female family members are likely to create stronger relationships with their
family members; however, female members have less power than male members in the
family (Barber & Haddock, 2003). Interestingly, even male children tend to have more
perceived power than the mother, grandmother, and sister. Additionally, daughters are
perceived to have the least power in the family, while fathers are perceived to have the
most. Therefore, it may be possible that the power difference may result in a loss of
intimacy between fathers and daughters. On the contrary, when mothers or grandmothers
take care of a male child, it may be difficult to supervise him due to the lack of perceived
Rosenblatt (1994) stated that, in the family, the mother-daughter relationship has
more boundary permeability than any other relationship because women are more willing
to express their feelings to others, and are more open to outside influence. Hence, Lemme
(2002) identified the mother-daughter bond as the strongest over time and as having the
Therefore, kin interaction normally follows the maternal side rather than the paternal side
7
of the family. In Kemp’s qualitative study (2007), the maternal grandmother was
perceived as a key person in obtaining social and emotional support. Likewise, in the
United Kingdom, maternal grandmothers have been identified as the most popular choice
grandmothers, and paternal grandfathers (Wheelock & Jones, 2002; Euler, Hoier, &
Rohde, 2001).
grandfathers seem to rely more on social support than grandmothers, who tend to utilize
family resources. This may be due to the fact that noncustodial grandfathers have less
interaction with family members and informal resources than noncustodial grandmothers.
According to Kolomer and McCallion (2005), custodial grandfathers had less symptoms
support systems, such as a spouse, owning a home, and working outside of the home.
Also, the study of Hayslip Jr., Henderson, and Shore (2003) identified that daycare
grandfathers are more likely to be involved in school, jobs, and finances while daycare
While daycare grandfathers have more social support, Bullock (2005) addressed daycare
grandfathers express powerlessness because they lack of control over their grandchildren.
8
While researchers suggest that there are differences between men and women in
terms of interacting with others, Hahlweg (2004) pointed out that the results of
longitudinal studies indicate no statistical gender differences. Hatch (2000) also stated
that gender is a much more complicated factor because it actively interacts with other
social entities. For instance, experience of social events, socioeconomic status, age, or
ethnicity may contribute to interactions among women and men. In other words,
tendencies of women and men in similar age groups and socioeconomic statuses may be
more similar than women among different age groups. In addition, cultural aspects of
gender influence the grandparent-grandchild interaction. Sandel, Cho, Miller, and Wang
(2006) mentioned that Taiwanese granddaughters tend to have less satisfaction and
closeness with their grandparents than do grandsons. They assumed that it may be a result
of the traditional Taiwanese cultural preference for sons. Another reason may be that girls
are more likely to be dissatisfied with the relationship if it is not as close as expected,
because girls are more emotionally sensitive. Therefore, it is unclear whether differences
The developmental stage of the child is another factor that affects the interaction
grandparents one century ago only knew their first grandchild; however, today some of
them see middle-aged grandchildren and reach great-grandparenthood. With this change
9
in situation has come a realization that the age of the grandchild may affect the
interaction between the first and the third generations of the family. Younger
grandparents tend to have closer relationships with grandchildren than older grandparents
(Lin & Harwood, 2003). In addition, younger grandparents tend to be more involved in
involved in caring for grandchildren because they are relatively physically healthy, and
their adult child has a young child who needs an intense level of supervision. In contrast,
as grandchildren grow older, grandparents tend to spend less time with them because
children may not need all-day supervision anymore and/or grandparents are losing their
physical health. Lemme (2002) stated that young grandchildren tend to be closer to their
grandparents than teenage grandchildren. Similarly, Strom and Strom (1997) found that
were perceived as more satisfied with their roles as grandparents than those with older
grandchildren.
Grandparent-parent relationship
Relationships between grandparents and parents of the child are another factor
mentioned that positive relationships between adult parents and their parents positively
affect children’s well-being. However, if grandparents and parents experience some kind
10
& Mutran, 2004). In general, older parents tend to have less ambivalence toward children
who are successful. Hence, older parents are more likely to feel ambivalence toward
unmarried adult children, those who have less education, and those to whom they provide
financial support (Peters, Hooker, & Zvonkovic, 2006). In this regard, grandparents feel
more stress in their relationships with adult children and their spouses who give custody
grandparents experience stress because adult children let grandparents take care of
grandchildren and are not responsible for childcare. Compared to custodial grandparents,
noncustodial grandparents are more likely to express concerns related to their husbands
rather than adult children, whether those grandparents are married or not.
Parent-child relationship
relationship. However, there has not been much research found to identify the effect of
relationship may be positive if the parent and the child are linked together. Goodman
referred to this pattern as the ‘child-linked family’. In this family type, the grandmother
has relatively low depression and high life satisfaction, and the grandchild also has a low
level of behavior problems, even though the grandmother-parent relationship is not linked.
On the other hand, in the ‘isolated-child family’, where the grandmother is linked to both
the grandchild and the parent of the child and there is no linkage between the parent and
the child, the child has the second-highest level of behavior problems after those in the
11
‘not-bonded family’. In this relationship, the grandchild may be hurt by poor
relationships with their parents even though the grandmother tries to take over the
parent’s role. Grandmothers in this family pattern also have a higher level of depression
and a lower level of life satisfaction than those in the ‘child-linked family’.
Culture
each culture perceives family and the role of grandparents. While Caucasian American
culture constitutes the majority in the U.S., African American families, Hispanic
American families, and Asian American families are the major three ethnic groups that
follow Caucasian American (Julian, McKenry, & McKelvey, 1994). Julian, McKenry,
people are basically equal regardless of age. Thus, Caucasian American people seek to be
independent from others, and this tendency continues to the senescence period (Lemme,
2002). Hence, older adults may live close to their adult children and interact with them
with others, Asian American, African American, and Hispanic American people tend to
pursue collateral relations with people (Gladding, 2002). Minuchin, Lee, and Simon
(2006) also stated that extended family networks are important for African Americans.
The extensive kin network of African American families provides both economic and
12
emotional support to their members (Becvar & Becvar, 2006). In the U.S., African
Americans have not historically had external resources; therefore, they are more likely to
help other family members than Caucasian American. Hence, African American older
adults are more likely to take grandchildren, nieces, or nephews into their home (Mitchell
& Register, 1984). According to Gladding, African American families have more open
roles, so they are less likely to stereotype based on age or gender. Hence, grandparents
tend to have a close relationship with their children and grandchildren. Also, they tend to
be less reluctant to become primary caregivers for their grandchildren. Hence, African
Caucasian Americans (Taylor, Washington, Artinian, & Lichtenberg, 2007). On the other
hand, African Americans feel responsibility to take care of their kin because this kin
family member, and the priority of family needs over the needs of individuals” (p.121).
This means men and older people have more power than women and younger people. In
traditional Latino families, the father has the authority in major family decision-making,
while the mother usually plays a traditional role as the wife. Also, the younger generation
has respect for the decisions made by the members of the older generation. Latino culture
also has collective characteristics since family is a priority. However, Becvar and Becvar
found that due to the variety within Latino culture, these characteristics may be difficult
to generalize.
13
According to Julian, McKenry, and McKelvey (1994), Asian culture also
emphasizes extended family function and respect toward elders, even though this varies
among ethnic groups. In general, the paternal side of Asian family relationships tend to
be stronger than the maternal side, which is due to male preference and Confucian ideas
(Lin & Harwood, 2003). For example, the majority of research participants in Lin and
Harwood’s study in Taiwan identified their lineage along the paternal side. Indian
families have similar characteristics as well. Raina (1989) stated that the father has
authority over the family, especially the son, in a classic Hindu family. Children are
required to obey the father and elders in the family; however, Raina mentioned that this
classic authority pattern is changing as children seek freedom from traditional customs.
The change occurring in India may also happen in other Asian cultures as individualism
becomes a more dominant ideology. For instance, Ikeda and Hatta (2001) described
characteristics of Japanese families, specifying that the family has no clear boundaries
between the mother and the child. The father tends to be isolated in the family system
because the expectation for men is to devote themselves to their business rather than to
build relationships with their families. However, Ikeda and Hatta also mentioned that
family structures in Japan are changing such that the father is more involved in the family
and shares relatively the same burden of childcare as the mother. The changes of family
14
Social support system
The social support system is another external factor influencing the grandparent-
grandchild interaction. The social support system may function as a third party in the
report physical, mental, and financial difficulties while caring for their grandchildren.
Custodial grandparents take over the custody of their grandchild because the parent of the
grandchild is not able to provide childcare. Hence, these grandparents are likely to sustain
a two-way relationship because the majority of them rarely receive support from a
grandchild’s parent. Therefore, the support from other family members or the community
may become a third party and complete the triangle. The social services available for
counseling services, and financial support (Sands & Goldberg-Glen, 2000). Gandparents
raising their grandchildren have expressed particular need for recreational activities for
the children, health services for both the grandchild and the grandparent, and parent
Consequences of Grandparenting
The well-being of grandparents who spend time with their grandchild may
frequency of contact with their grandchild, and closeness with their grandchild may
influence their well-being, in terms of degree of depression and life satisfaction. There
are previous studies that illustrate custodial grandparents’ health conditions compared to
15
non-custodial grandparents. According to Hayslip Jr. and Kaminski (2005), custodial
insomnia. Hayslip Jr. and Kaminski also stated that difficulty of performing daily
activities appears more frequently among grandparents raising grandchildren than those
who do not care for their grandchildren. Robinson and Wilks (2006) mentioned that one-
third of participants in a study by Minkler, Roe, and Price (1992) expressed that their
grandmothers in the study stated that they limit their activities with grandchildren due to
their health, and they also have limited time for themselves, they still say that they are
willing to take care of their grandchildren if the children’s parents are not able to do so
experience mental health issues while they are serving as caregivers for their
grandchildren. In Ross and Aday’s study (2006), 94% of their research participants,
while parenting their grandchildren. McGowen and Ladd (2006) conducted an on-line
groups, the custodial grandmother group had a significantly lower score in life
satisfaction and success as grandparents. It is assumed that the lower score of the
16
custodial grandmothers may result from their increased responsibility as a caregiver,
Although grandparents may feel they have less time to be with their age group
after becoming a childcare provider for their grandchild, there are rewards for
grandparents. One of the benefits is financial rewards from the adult child (Presser, 1989).
Presser found that 31.1% of daycare grandmothers received cash payments for providing
childcare, and 11.7% receive non-cash payments, including meals and transportation. In a
study by Wheelock and Jones (2002) conducted in the United Kingdom, one-third of
parents gave in-kind benefits to daycare grandparents even though they rarely gave cash
benefits. In addition to the financial rewards, Wheelock and Jones also found that
grandparenting gave special moments to grandfathers who had, for the first time, the
revealed in the Silversteins’ study (2007). Emotional satisfaction also comes from the
feeling of taking on a social career. Even though daycare grandparents do not receive
financial rewards, they often consider childcare an occupation that positively impacts the
On the other hand, childcare is stressful for both custodial and noncustodial
grandparents. Goodfellow and Laverty (2003) found that caring for grandchildren was a
17
physically and emotionally exhausting job for both custodial and noncustodial
disempowerment and family obligation. McGowen, Ladd, and Strom (2006) stated that
scores than custodial grandparents. Lack of free time is another factor that brings stress to
custodial and noncustodial grandparents. Goodfellow and Laverty found that even though
both custodial and noncustodial grandparents wanted to spend time with their peers, it
For parents and grandchildren, on the other hand, grandparents become a great
support system as they enter into the triangle relationship. Specifically, parents of the
child may benefit most since the grandparents’ main role is to help care for the child.
Informal child care from grandparents reduces the financial burden for the parent (Presser,
1989: Wheelock & Jones, 2002). In addition, it also reduces the social costs for providing
benefits, parents prefer grandparents as caregivers because they have the same family
values (Wheelock & Jones, 2002). For instance, grandparents are able to educate
grandchildren within the culture of the family, especially when the grandchild is raised in
a different country from where the grandparents and parents were raised.
18
Besides the benefit to the parent, grandchildren also benefit from the triangular
relationship because they have more time under the adult’s guidance. Silverstein and
transmission of the mother’s depression to the children, although they do not influence
grandchildren’s well-being indirectly and in the long term. In addition, researchers found
19
Chapter 3: Theoretical Frameworks
Key concepts
& Kiyak, 2005). Compared to Erikson’s developmental stages, which assume that the
individual’s life stage is set up from birth, the life course perspective refers to an
gender, income, and social class. In other words, the life course perspective admits the
diversity of the individual’s roles, which change throughout the individual’s life span.
Hence, the life course perspective accepts multiple social roles throughout the life span
(Macmillan & Copher, 2005). Because it accepts the heterogeneity of an individual’s life
pattern, grandparenting is also considered as a part of the individual’s life span. However,
skills (Dolbin-MacNab, 2006). Moreover, the life course perspective describes different
life patterns among individuals with different backgrounds. Thus, the life course
perspective may be helpful to address how socioeconomic status affects the use of
The life course perspective defines three types of time: individual, historical, and
generational. First of all, individual time refers to chronological time, which focuses on
20
time describes culture-appropriate roles, obligations, and privileges for individuals within
time recognizes appropriate norms within society, Clausen also called this social time.
Hence, the laws of society dictate a person’s rights and duties based on the person’s age.
For instance, a six-year-old child in the United States is expected to attend school. As he
approaches 16, the child will have a right to drive but does not obtain the right to drink
until the age of 21. Similarly to the child in the U.S., a child in Korea begins schooling at
the age of 6. However, the child in Korea will have the right to drive at the age of 18 and
the right to drink at the age of 19. While social time mostly follows chronological age, it
is not fixed to it (Clausen, 1986). There are cases where informal expectations, rather
than formal regulations, decide the age-appropriate norms. For example, some students
get into the workforce right after they finish high school, while others continue to remain
at school for undergraduate and graduate study at college. Another example is the
marriage age: some couples get married in their early 20s, while others get married in
Secondly, the cohort effect is related to another type of time, historical time, as
well. Historical time emphasizes societal changes over history and how societal changes
affect individuals (Price, McKenry, & Murphy, 2000). In other words, historical events,
communicate with their friends using text messages, e-mails, and on-line messengers,
21
while some adults still have a hard time using a computer. Thus, interaction within a
family is often influenced by how well family members are able to use the Internet and
other technologies. For instance, before advancements in technologies, when a child was
out of the country for military service, it would take weeks to exchange letters. But these
days, a child in the service is able to chat online with family members at his or her
convenience.
Lastly, generational time refers to the positions of individuals within their family
and their roles and identities related to their positions (Price, McKenry, & Murphy, 2000).
It is somewhat similar to individual time, which defines individual roles according to age.
However, generational time focuses more on the role within family. Hence, a 46-year-old
grandmother and a 98-year-old grandmother are in different age groups, yet they are both
Nowadays, people can play the role of child for approximately 50 years and maintain
that only 3% of women over 60 years old had at least one living parent in the 1800s.
Comparatively, more than 60% of people in this age group had at least one living parent
in 1980. Since it has become common for two or more generations to share lifetimes
much longer than in the past, unexpected off-time, such as early parental death, may
result in personal crisis because the individuals are unprepared and have limited support
from peers.
22
Strengths of the life course perspective
The life course perspective enables one to see individuals both from their
& Murphy, 2000; Hareven, 1994). Hence, it is possible to understand roles, rights, and
responsibilities of individuals from the perspective of both the individual and the family.
In other words, the life course perspective perceives the individual’s life transition as
relating to family transitions (Hareven, 1994). That is, changes in the individual’s life
results in changes in the family structure. Hence, not only has the role of grandparents
caring for their grandchildren changed from the grandparent to the caregiver, but also the
family structure has changed from the parent-headed family to the grandparent-headed
family.
same way, assuming that people in a certain stage should have the same role and same
tasks; however, the life course perspective allows individuals to set their own roles and
tasks (Hooyman & Kiyak, 2005). According to Erikson’s theory, parenting is the main
task for middle adulthood. Although grandparents raising grandchildren are relatively
young, and some may be still in middle adulthood, a second round of caring may not be
their expected task, based on Erikson’s theory (Anderson, Carter, & Lowe, 1999).
Consequently, grandparents who have to deal with two life tasks at the same time –
parenting grandchildren and dealing with their own life tasks – may fail to establish a
plan for their future. Hence, maintaining two roles at the same time may result in role
23
confusion and caregiving stress. Even though Erikson’s theory explains why
grandparents raising grandchildren suffer from role confusion and isolation from their
peers, it does not identify whether grandparent caregivers are in middle or late adulthood.
As only individuals with children are able to complete the sixth and seventh stage,
In contrast, the life course perspective accepts both 40-year-old individuals and
the grandparent. In other words, the life course perspective describes different life
patterns among individuals with different backgrounds (Price, McKenry, & Murphy,
2000). Hence, the life course perspective accepts multiple social roles throughout the
life pattern, it also considers grandparenting as a part of the individual’s life span
socioeconomic status affects the use of support systems and the well-being of the
grandparent.
In addition, the life course perspective focuses not only on families with married
parents but also other types of families, such as divorced families, remarried families, and
families in poverty. The divorced family life cycle starts from the decision to divorce and
goes through the separation period and the final divorce phase (Price, McKenry, &
Murphy, 2000). In contrast, the remarried family life cycle has three phases: recovery
from loss and entering a new relationship, planning a new marriage, and family
reconstruction. In addition, the life course perspective admits different family cycles
24
according to the family’s financial situation. For instance, families in poverty tend to
have less calendar time because they spend more time overcoming economic disparity;
thus, it is harder for individuals in these families to accomplish each stage’s task due to
the shorter calendar. Recently-begun studies on these family formations may differentiate
the life course perspective from stage theories, which do not include the diversity of the
type of family.
The life course perspective has limitations in explaining individuals within the
family. First of all, it inadequately accounts for stress occurred from role overload and
taking a role off-time. For instance, it may not be able to identify the cause of stress that
custodial grandparents often experience in terms of role confusion and role overload. The
life course perspective is similar to role theory in that it allows individuals to take
multiple roles and responsibilities at the same time. In addition, both life course
perspective and role theory explain role changes and role loss throughout the life span.
For instance, in role theory, the role is formulized based on the discussion of the
socialization and adjustment of the individuals (Biddle, 1979). In other words, role theory
is similar to the life course perspective in that both emphasize family interactions. Hence,
from initial socialization in the family, role theory perceives that individuals learn role
behaviors that are appropriate for their social class, ethnic group, and other social
25
Compared to the life course perspective, however, role theory describes role
confusion and role overload of individuals. As the child develops his or her roles through
role playing and role taking, adults also form roles (Hooyman & Kiyak, 2005). On the
other hand, people may be confronted with role conflict during the process of role
formation. This means that individuals are sometimes required to perform two or more
tasks even though they are not able to do all of them. Or, people may sometimes suffer
from role overload when they are asked to do too much. Because the life course
perspective permits individuals to perform multiple roles at the same time, it is also
possible for grandparents to obtain a new role as a caregiver for their grandchildren when
the birth parent of these children is working during the day. In this situation, the life
course perspective may recommend that the grandparents and the parent should maintain
a balance in their roles of parenting; however, it may not describe what would happen
The life course perspective and role theory have different perspectives on the
timing of when an individual takes on a particular role. In the life course perspective,
generational time allows individuals to take on the same roles at a different individual
time. Role theory also describes individuals who assume a role at an off time. However,
the life course perspective does not explain the possible consequences of taking a role at
an off time, while role theory considers this an ambiguity of formatting culturally
appropriate roles (Landry-Meyer & Newman, 2004). For instance, grandparent roles are
26
family. In addition, those who become a grandparent in their 30’s may have different
theory. It means that the life course perspective has not been verified for diverse ethnic
groups and diverse cultures. The life course perspective is primarily based on white male
individuals. Because each culture has different perceptions of time, individuals from
different cultures may have different individual, historical, and generational time. In
addition, comparisons among people who live in the same location but have different
cultural backgrounds may also need to be identified in terms of their life course experiences.
It means there is heterogeneity within the same cohort. Different life courses are possible
among members of the same cohort even after they experienced the same historical event.
Also, due to the limitation of a life course ‘perspective’, it may be less able to predict
various social phenomena. Hence, researchers need to collect data from those who have
Key concepts
Bowen family systems theory defines family as a system that gives and takes
energy from each of its members. In other words, systems continuously move between
change and stability. An individual in the family may sometimes get into an unstable
situation when tension exists within family. If so, change occurs until the situation
becomes more stable. Then, stability will be maintained unless it falls into another
27
unstable position that will need feedback to help it move back to a stable status (Becvar
& Becvar, 2009). Bowen family systems theory has eight key concepts: (1)
differentiation of self, (2) the triangle, (3) nuclear family emotional process, (4) family
projection process, (5) the multigenerational transmission process, (6) sibling position, (7)
emotional cutoff, and (8) emotional processes in society (Papero, 1990; Nichols &
Schwartz, 2001).
(Papero, 1990). Nichols and Schwartz (2001) mentioned that differentiation refers to the
ability to act flexibly and wisely even in an anxious situation. Hence, differentiated
people tend to possess strong emotions but they are able to resist the pull of emotional
others. Papero defined emotional as “forces or pressures deeply rooted in each individual
and between the individual and his/her environment” (p. 45). In this sense, the emotional
system refers to “the reactivity of the individual to its environment” (p. 45). Papero
mentioned that the basic level of differentiation is usually established in early life,
although it may be modified in later life as well. However, once the basic level of
The triangle relationship can be observed when anxiety within the system is too
intense (Papero, 1990; Prest & Protinsky, 1993). The anxiety in the family system
evolves when the interpersonal needs of the family members are challenged. In Bowen
third person may be able to join the twosome interaction and make a threesome
28
relationship (Ackerman, 1984). Ackerman assumed that there is no simple one-on-one
interaction among human beings. Rather, a third thing or individual constantly interrupts
an anxious twosome relationship in order to stabilize the relationship. Papero also stated
that the triangle is so natural in everyday life that people are often unaware of it. Griff
the family intervention positively influenced the treatment of the grandchildren. However,
the third person in the triangle is not necessarily a part of the family. For example, in a
single parent family with a child, although a grandparent may be involved in an anxious
situation with the parent, a child’s teacher or a friend of the parent may become the third
person. In order to function positively, Ackerman stated that each relationship should
maintain certain distinctions from the others, and each person in a relationship should
have direct interactions with the other. In addition, the distance between persons should
be at an equilibrium, which means it should not be too close or too far away. On the other
hand, Bowen stated that the anxiety of the third party would increase as he or she became
more involved in the triangulation (Miller, Anderson, & Keala, 2004). When the anxiety
evolves in the marital relationship, a child often becomes the third party. Bowen was
concerned that the third party child would have emotional distress and negative social
outcomes.
The nuclear family emotional process is related to the level of a common self,
“we-ness” between couples (Papero, 1990, p. 51). However, the emotional process in the
nuclear family is often related to the extended family system because each partner applies
mechanisms that he or she used in relationships with his or her parents (Titelman, 1998).
29
The outcome of their efforts to develop the common self is a result of four mechanisms in
the nuclear family: emotional distance, marital conflict, transmission of the problem to a
child, and dysfunction in a spouse (Papero, 1990; Nichols & Schwartz, 2001). These four
mechanisms may be used in a combination of two or more at the same time. Emotional
distance may mean either external or internal distance. Couples with external distance
have less time together because they are physically apart from each other. In another
sense, internal distance often occurs without the members of a relationship being aware
of it. Papero stated that couples experiencing internal distance gradually talk less and less
with one another and finally have nothing to share with each other. Marital conflict is one
because couples with conflict are likely to be more distant. Subsequently, couples with
great internal or external distance are likely to have marital conflict. When couples have
marital conflict, it affects their children as well because all children are likely to be
Spira and Wall (2006), anxiety between grandparents and parents may influence
partners. In a marital relationship, couples compromise with each other in order to avoid
a conflict situation. In this relationship, each person has roughly the same amount of
responsibilities. However, if one person is under-functioning and the other person is over-
functioning, the family may experience conflict when an anxious situation arises.
30
According to Papero (1990), the family projection process relates to the
relationship between a mother and a child. In other words, the family projection process
focuses more on the mother than the father, which works under the assumption that the
mother’s emotional sensitivity to a child is greater than that of her spouse. The role of the
husband is to support the wife’s interaction with the child, while the father is equally
involved in the attachment process with the child. Hence, Bowen family systems theory
assumes the mother to be a primary person for the child. Thus, the father’s functioning
impacts the relationship of the mother and the child, not the father and the child. If the
father does not exist, the intensity between mother and child increases, whereas in a
family with the father’s support, the relationship between mother and child is more stable.
process in terms of describing the effect of parents on their children (Papero, 1990).
parent-child involvement on the child’s development. In other words, the focused child to
whom the parent pays more attention is likely to be more attached to the parents, and
tends to be more vulnerable to emotional intensity. On the other hand, the less-focused
child is more differentiated from the parents, so the child is likely to learn more from the
parents’ strengths. Bowen proposed that people marry partners who possess a similar
level of differentiation. Therefore, focused children are more likely to seek partners with
similar or less differentiation, whereas less focused children tend to marry partners with
31
The sibling position describes characteristics of each child based on his or her
position within the family (Papero, 1990). In other words, Bowen assumed that the level
perceived that children’s personalities are developed based on their position in the family
(Nichols & Schwartz, 2001). Hence, the youngest son with sisters tends to show different
characteristics from the youngest son with brothers only. Nichols and Schwartz said that
firstborn children tend to apply power and authority to defend their status from other
siblings. In contrast, later-born children are more likely to be open to exploring because
attachment to their parents and other individuals (Papero, 1990). In other words, it refers
to the way people control anxiety between generations (Nichols & Schwartz, 2001).
Papero stated that individuals can maintain emotional distance using either psychological
emotional cutoff, he or she can be isolated from a family member even though he or she
lives with them. On the other hand, with physical distance, an individual can control the
frequency of his or her contact with parents. For instance, people may control emotional
attachment by reducing the number of phone calls or visits. Or, people simply live away
processes in the family (Papero, 1990). Individuals’ anxieties tend to enhance functioning
as two people add a third person to their group. Similarly, in terms of emotional
32
processes in society, as the level of anxiety increases, society tends to seek equilibrium
by composing more stable groups. This often happens where one entity is added as a
Anderson, Carter, and Lowe (1999) mentioned that systems theory permits the
observer to investigate different levels within the same phenomenon at the same time and
to analyze the whole as well as the parts. Although it may make analysis more complex,
it is likely to be more accurate and closer to real life. Therefore, based on systems theory,
social workers try to observe not only clients’ personal characteristics but also their
clients’ environments, it is possible to more thoroughly analyze both their problems and
their strengths that may lead to possible solutions. Systems theory treats the client’s
environment as containing influential factors that affect assessing, and intervening in, the
client’s problem. Systems theory’s influence on the social work field is evident in the
current common practice of drawing genograms and eco-maps for assessment, and
Compared to the life course perspective, family systems theory focuses more on
Bermann, 2000). As family systems theory studies both micro and macro perspectives,
such as communication, transactions, and anxiety in the family, it is a better tool for
understanding behaviors of each family member than the life course perspective.
33
Specifically, because Bowen family systems theory emphasizes the triangle relationship,
it is able to show a solution for role overload. The life course perspective has a weakness
in explaining role overload; however, Bowen family systems theory provides a triangle as
a solution when family conflict occurs. For example, if a family member’s role overload
is the reason for anxiousness within the family, the family may be able to reduce the
person’s overload by adding a new person who is able to take part of the load.
Moreover, Bowen family systems theory is good for describing the relationship
the extended family system. Although Bowen named the nuclear family’s emotional
system as one of the major concepts of family systems theory, the family as system refers
to at least three generations (Titelman, 1998). The nuclear family consists of the mother,
the father, and the child. In addition to that, the father’s family of origin constitutes
another emotional system, and the mother’s family of origin yet another. Hence, Nichols
and Schwartz (2001) stated that Bowenian therapy begins with collecting descriptions
and histories of existing problems by assessing the extended family systems. Bowenian
therapists assess whether the problem is related to events in the extended family. Today,
the majority of families are nuclear families that only include parents and their children.
many ways. For instance, grandparents are often involved in the families of their adult
children in order to support the family. Specifically, custodial grandparents have been
involved in parenting their grandchildren when parents are not able to provide care for
their children. The triangle may exist in families with non-custodial grandparents as well.
34
Noncustodial grandparents may sometimes become part-time caregivers while parents are
at work. Hence, Bowen family systems theory is particularly strong in illustrating the
changes happen constantly within and between the systems. Gerson (1996) stated that
system changes that are designed to address one problem may accelerate another problem.
Bowen stated that the anxiety of a third party entering in an anxious twosome relationship
would increase although the anxiety among the other two parties would decrease once the
third party joined (Becvar & Becvar, 2009). For instance, if a grandmother becomes a
primary caregiver for the grandchild due to a parent’s drug problems, the child may
obtain consistent care from the grandmother. Although grandparenting may benefit the
child, it may result in negative consequences for the grandmother who provides care for a
second time around. While Bowen stated that the anxiety of the third person, after being a
correlations between the anxiety of the third party and triangulation. In other words, some
relationship resulted in a negative attitude toward family. On the other hand, other
researchers stated that the impact of triangulation and emotional distress were not
significantly related to each other (Miller, Anderson, & Keala, 2004). In this aspect,
therapists are constantly challenged to determine whether the client’s system needs to be
35
changed or not. If they decide to make changes, the concern becomes which system,
among the various systems that exist in the client’s situation, should be utilized, because
Another weakness of Bowen family systems theory is that it focuses too much on
family conflict and anxious situations as causes of people’s mental health problems.
Bowen’s theory was originally derived from family therapy, especially while he was
working with people with mental health problems such as schizophrenia. Bowen stated
that a high level of anxiety in the family may cause mental health issues for an individual
in the family (Becvar & Becvar, 2009). However, it may be rare that the family has no
conflict at all. Also, it is true that the family is a fundamental system that strongly affects
people’s physical and mental conditions. On the other hand, as the family is one of many
systems, there are factors besides family conflict that influence people’s psychological
dysfunction.
Lastly, gender bias exists in Bowen family systems theory. When explaining the
emotional process of the family, Bowen focused on the interaction between mother and
child. Although the mother has traditionally been a primary caregiver for children, this
situation is currently changing as more women enter the workforce. Bowen family
systems theory is still based on traditional mother-father roles, in which the mother stays
at home caring for the children while the father works outside of the home. Even though
Bowen family systems theory covers extended family and blended family structures, its
basic assumption of gender roles may be a weakness when it comes to explaining diverse
types of families.
36
Life Course Perspective and Bowen Family Systems Theory
Heterogeneities
The main differences between the life course perspective and Bowen family
systems theory are the considerations of diversity and time. Regarding diversity, the life
course perspective assumes that the individual’s life course varies according to cohort,
gender, and race (George & Gold, 1991). George and Gold also stated that scholars have
found that the life course is “less predictable and more heterogeneous than was initially
imagined” (p.71). In the life course perspective, there is variability across cohorts, and
even among the same cohort, based on family characteristics. Hence, rather than
analyzing each individual’s life course with established norms, the life course perspective
tends to recognize the diversity of each person’s life course. Thus, in the life course
perspective, taking roles off time is considered as a pattern of the life course, not as a
problematic situation.
Although Bowen family systems theory also perceives that each family has
unique subsystems in the family, it tends to find norms and similarities within family
systems. Bowen family systems theory assumes that the anxiety in the relationship of two
people may be mediated by the third party (Miller, Anderson, & Keala, 2004). However,
Bowen assumes the distress of the third person may be increased as he or she plays a
mediator. This assumption is applied to all families, whether or not they have different
cultural backgrounds. Therefore, when anxiety exists in the marital relationship, a child
often takes a third party role to reduce anxiety. In addition, a single parent with a child
37
may need a third person, such as a grandparent or an adult sibling, to avoid emotional
distress. On the other hand, the life course perspective focuses on the changing roles of
the parent and the child in a given time period. Hence, the triangle relationship may not
be necessary because each individual has his or her own shifting roles.
two theories. As its name represents, the life course perspective is based on the
individual’s life course. Therefore, life course analysis needs a dynamic and longitudinal
perspective (George & Gold, 1991). This aspect may be both a strength and a weakness
of the life course perspective. It may be a strong theoretical framework because the life
course perspective includes changes in people’s life from birth to death within a certain
society. However, it is hard to collect such data to analyze trajectories; therefore, George
and Gold mentioned that most of the life course perspective research examines transitions
Bowen family systems theory, however, focuses more on the current system of
the family rather than the pathways of people’s lives. Hence, compared to the life course
perspective practitioners, Bowenian therapists are more likely to assess the current
problem within the family and its effects on family members. They also examine family
history in order to identify the problems existing in the family; however, Bowen family
systems theory tends to focus more on the system of the family itself rather than the
38
Reciprocities
Both the life course perspective and Bowen family systems theory emphasize
interactions within the family. Macmillan and Copher (2005) stated that role pathways in
the life course perspective illustrate the cause and the effect of interdependent family
roles. Putney and Bengtson (2005) mentioned that the life course perspective concerns
the same sense, family is the main place where people learn different types of roles and
how people assume and change roles. Therefore, a role change of a person in the family
affects the roles of other people in the family. For example, when parents are not
available, the eldest sibling often becomes a parental child who is responsible for caring
for younger siblings. The eldest sibling may prepare meals, help with homework, or
grandparent who takes a parental role for the grandchild when the parent is not able to
take care of the child. The life course perspective perceives grandparenting as acceptable
when the family needs it. Hence, taking on either full-time or part-time caregiving roles
Similarly, Bowen family systems theory also concerns interactions among family
members. As in the life course perspective, Bowen family systems theory accepts the
grandparent entering into the relationship of the parent and the child. Bowen family
often apply a genogram. Nichols and Schwartz (2001) stated that genograms are
39
“schematic diagrams listing family members and their relationships to one another” (p.
85). The genogram includes ages, marital status, deaths in the family, and gender of each
family member. It is a static dyad because it includes the conditions of the relationships
among listed people, such as conflicts, strong bonds, cutoff, or triangles. Bowen family
systems theory assumes that family interactions include both the cause and the solution of
therapists may illustrate supportive factors and risk factors within the family.
support for family members, kin interactions, in some cases, are considered more
generational ties are more ambivalent than horizontal sibling-ties or friend-ties. Moreover,
the life course perspective holds that historical events and cohort effects function not to
eliminate this ambivalence, but to set it at the societal level. Hence, older parents may
have both positive and negative feelings when they interact with their adult children. In
the qualitative study of Peters, Hooker and Zvonkovic, older parents generally wished for
more frequent contact with their adult children; however, they were concerned about
future requests for help from their children as well. Similarly, adult children expressed
their ambivalence toward their older parents. Although they were willing to help their
parents, they were concerned about potential caregiving. Bowen family systems theory
has similar perceptions to those of the life course perspective. Bowenian therapists focus
40
Besides internal interactions, Macmillan and Copher (2005) stated that family
roles involve complex interdependent relationships with extra-family roles in the life
time and historical time. Social events that occurred at a certain time influence people’s
roles in the family. For example, when economic crisis occurred in Korea a decade ago,
employment became unstable. Hence, women who had stayed at home started to come to
the workforce to support the family finances. When both parents entered the workforce,
parenting roles in the family changed as well. In the past, women were mainly
responsible for caring for the child; however, as women have acquired worker roles,
grandparents have often taken parent roles while parents are at work. Putney and
Bengtson (2005) described the extra-family effects that the mother’s entry into the
workforce may enhance, such as the family’s economic wealth; but, it also may result in
changes in the family routine or the balance of the mother’s work with housework.
individuals and their environment. An ecomap is applied to address the interactions that
occur with external factors. While a genogram illustrates interactions in the family, the
ecomap describes interactions with social institutions outside of the family, such as
community, religious groups, schools, workplace, and so on. The ecomap is helpful if one
aids in assessing whether any of the extra-family factors might help to address the
individual’s problem.
41
Rationale
grandchildren and its relation to the grandparents’ well-being within the context of the
family relationship. Previous studies described in the literature review section have
examined gender, age, ethnicity, and/or legal guardianship of grandparents and its impact
the context of the family system. In other words, addressing the grandparent-grandchild
relationship is important because the grandparent, grandchild, and parent often make the
triangle of the family system. This triangle may influence both positively and negatively
the grandparent’s well-being as grandparents become a third party of the system. Hence,
this study focused on this grandparent-grandchild relationship and its impact on the
grandparent’s well-being. Then, this study compared the interaction between the
grandparent’s well-being and the closeness between the grandparent and the grandchild
The closeness between the grandparent and the grandchild may depend on various
well-being may be related to other factors besides the relationship between the
grandparent and the grandchild. Hence, it is necessary to examine factors associated with
these factors, some grandparents may feel reluctant to take care of the grandchild, while
42
multigenerational families are more common in non-Caucasian American culture;
role more easily than Caucasian American grandparents. Also, grandparents whose
relationship with their adult child is troubled tend to be more reluctant to take care of
their grandchildren than those who have a good relationship with their adult child (Welsh
& Stewart, 1995). Thus, in this study, three factors – grandparent’s functional capacity,
grandparent-parent support, and cultural attitude – were included to address whether they
affect the interaction between the grandparent’s well-being and the grandparent-
grandparents have a closer relationship with their grandchild; who has a higher level of
well-being; and how a grandparent’s functional capacity affects the grandparent’s well-
being.
Lastly, it is possible to investigate how Bowen family systems theory and the life
course perspective can be applied to the grandparent’s well-being and the closeness
between the grandparent and the grandchild. Bowen family systems theory emphasizes
typically utilize a formal public day care center to meet their childcare needs; however, in
some cases, a grandparent may become the third part of the triangle by providing
informal private care. Grandparents have several strengths as caregivers. They are
flexible, so parents do not have to pick up their children at certain time. Also,
grandparents are able to provide one-on-one guidance to the child that is not available at
a childcare center. Moreover, grandparents are able to teach family values to their
43
grandchildren. Thus, both parents and children may have more stable relationships as
address the changes in the grandparents’ well-being as they spend time with their
grandchildren. Bowen family system theory addresses the third person’s anxiety after he
or she becomes part of the triangulated relationship. Thus, the amount of time spent with
grandchildren and emotional ties with grandchildren may influence the grandparent’s
well-being.
According to the life course perspective, a person is able to function in more than
two roles at the same time, and the roles are changeable as the family structure changes.
and caregiver, at the same time. Grandparents are able to continue to contribute to society
by providing childcare services to their adult children and grandchildren. At the same
time, they maintain the role of a traditional grandparent since they do not have legal
in family interaction between younger grandparents and older grandparents may affect
Research Questions
1) Do the age of both grandparents and grandchildren and the number of grandchildren
in the family affect the number of nights spent with grandchild and the closeness
44
2) Are the number of nights spent with grandchild and the closeness between the
being?
male or female?
custodial or noncustodial?
45
Chapter 4: Research Methods
This study is a secondary data analysis using the third wave of the National
Survey of Families and Households (NSFH) dataset. Of the main respondents of the
dataset, only those who had at least one grandchild in the family were included in the
study. The researcher analyzed the study population using the Structural Equation
Modeling (SEM) method. Latent variables included in this study were: (1) number of
nights spent with the grandchild; (2) closeness between the grandparent and the
both grandparents’ and grandchildren’s age and the number of grandchildren in the
family were included. Three multigroup comparisons were also implemented to identify
Dataset
The data set analyzed in this study was from the National Survey of Families and
Households (NSFH). The NSFH was a national representative panel study conducted by
the University of Wisconsin Survey Center for professors James Sweet and Larry
46
Bumpass of the Department of Sociology at the University of Wisconsin-Madison. The
initial interview with main respondents was conducted between 1987 and 1988, and wave
two was conducted from 1992 to 1994. Data for wave three, used in this study, were
collected between 2001 and 2003 via telephone survey, as were the original two waves.
The average number of call attempts for those who completed the survey was twelve, and
seventeen for those who did not complete the survey. Approximately 50% of those who
completed finished their survey in 5 call attempts. The final sample size in wave three
was 9,230, which was equivalent to 72% of those located (Sweet & Bumpass, 2002). The
Sampling methods used in the NSFH was a national multistage area probability
sampling that contained approximately 17,000 housing units from 100 sampling areas
across the nation. For the oversample, the researchers doubled the number of households
selected in the 100 sampling areas. Main respondents were originally selected from the
national sample of 13,007 people, including a main cross section of 9,637 households
with an over sampling of African Americans, Puerto Ricans, Mexican Americans, single
parent families, families with step children, cohabiting couples, and recently married
persons. Then, one person per household was randomly selected as the main respondent
(Sweet, Bumpass, & Call, 1988). The main respondent should be 19 years old or older to
be eligible to interview. However, those who were currently married were considered as
the main respondent although they might be under 19. In addition, if there was no family
member over 19, people under 19 years old were considered as eligible main respondents.
Children were classified from child 1 to child 7 according to their relationship with the
47
main respondent and living arrangements. In addition, 924 proxy interviews from a
family member of the main respondent were conducted in the wave three when the main
A primary reason for using this dataset for this study is that the NSFH has
variables the researcher would like to analyze. First of all, there are three generation
variables. When the main respondents are set to the first generation, not only does the
NSFH include variables of the first generation, but it also incorporates the variables of
the main respondents’ children and grandchildren. The dataset collects a particularly wide
range of information on the child of the main respondent. This information will allow the
researcher to identify the relationship between the main respondents, the first generation
and the child, and the second generation, as well as general characteristics of the second
generation. Thus, the NSFH is the most appropriate tool to facilitate the main goal of this
study, which is to examine the relationship among three generations in relation to the first
generation’s well-being.
Secondly, the NSFH enables the researcher to analyze the impact of the closeness
between the grandparent and the grandchild on the grandparents’ well-being. It has
variables measuring the closeness between the grandparent and the grandchild. In
addition, the dataset has several well-being measures, including depression, happiness,
and life satisfaction, which the researcher wants to measure. Hence, this dataset allows
the researcher to determine how the closeness of the grandparent and the grandchild
48
Another benefit of the NSFH is that it includes a large national sample across the
nation. More than 7,000 people participated in the research, and more than 25% of
grandparents who are from various age groups and diverse cultural backgrounds.
A limitation of using the NSFH dataset is that there are few variables measuring
information, such as exact age and gender of each grandchild. In addition, it would be
beneficial if there were a variable addressing whether the grandchild is from the
daughters’ family or from the sons’ family. Moreover, the dataset has only two variables
regarding grandchildren: the age of the youngest grandchild and the age of the oldest
characteristics.
Sample
The sample in this study was selected from the NSFH wave three dataset. While
the total sample size of the wave three was 9,230, the sample in this study was limited to
the main respondents. In other words, data from the main respondents’ spouses, partners,
sons, and daughters were excluded in this study. Among the 7,277 main interview
respondents, the final sample used in this study consisted of those who answered that
they had at least one grandchild. The variable, RH9, asked how many grandchildren the
respondent had. The researcher filtered out cases that responded to the RH9 with
‘Refused to answer’, ‘Don’t know’, or ‘None’. After the researcher filtered out
49
nongrandparent cases and missing cases, the sample size decreased to 2,541. At the time
of data collection (2003), the mean age of grandparents was approximately 56 years old.
In addition, the youngest grandparent was 36 years old, and the oldest was 105 years old.
It was unnecessary to consider sample weights in this study because the sample
from wave three was no longer a national probability sample. In the wave three, only a
subset of the original sample was interviewed. The subset consisted of main respondents
who were 45 years old and older by January 1, 2001 with no focal children; respondents
who had a focal child aged between 18 years old and 33 years old; spouses or partners of
main respondents in the wave one; and respondents’ eligible focal children who were
between 18 and 33 years old in the wave three. The focal child was randomly selected
among those who were between 13 and 33 years old at wave three no matter whether or
Based on the research questions stated in the previous section, six research
hypotheses emerged:
1) The age of both grandparents and grandchildren and the number of grandchildren in
the family affect the number of nights spent with grandchild and the closeness
2) The number of nights spent with grandchild and the closeness between the
grandparent and the grandchild are associated with the grandparent’s well-being.
50
3) The grandparent’s functional capacity, grandparent-parent support, and cultural
male or female.
custodial or noncustodial.
Measures of constructs
Six latent variables were included in the study: (1) number of nights spent with
grandchild; (2) closeness between the grandparent and the grandchild; (3) grandparent’s
wellbeing, (4) cultural attitude; (5) grandparent-parent support; and (6) grandparent’s
functional capacity.
Two latent independent variables were separated after the initial analysis, which
indicated that the number of nights spent with the grandchild and the other three observed
variables contradicted each other in terms of describing the relationship with the
grandparent’s well-being. In other words, the number of nights spent with the grandchild
tended to negatively affect the grandparent’s well-being; however, the other three
check also supported two latent independent variables. Cronbach’s alpha was only .055
51
when calculated including all four observed variables as one latent independent variable.
However, when the number of nights spent with the grandchild was separated from the
other three variables, the alpha level increased to .697. Among the four variables, nights
intended to quantitatively measure the closeness of the grandparent and the grandchild.
On the other hand, the grandparent’s self-rating of closeness was intended to measure the
the grandparent’s well-being. Depression and life satisfaction scales have traditionally
been popular methods of measuring older adults’ well-being in researches. In this study,
well as depression and life satisfaction because the correlation results of four scales were
all over .40. In addition, the researcher assumed that four different instruments would
analyze the well-being of the grandparent from a more diverse aspect than using fewer
instruments.
Three indicators of cultural attitude used in this study were from a set of cultural
attitude scales. Among them, the researcher chose five variables that were related to
explain the three generational relationships represented in this research. Among those
five, three indicators were ultimately included to measure cultural attitude based on a
reliability test. The functional capacity variable and the grandparent-parent support
52
Observed variables
Several observed variables were used in this study. For instance, three variables,
the age of the grandparent, the age of the grandchild, and the number of grandchildren in
the family, were included for testing hypothesis one. The age of the grandchild is
measured through two variables, the age of the oldest grandchild and the age of the
Although gender and legal custody of grandparents were from the wave three data, the
ethnicity variable was from the wave one dataset. Thus, the researcher merged the wave
three dataset with the wave one dataset in order to include the respondents’ ethnic
backgrounds.
Latent independent variables: Nights spent with grandchild and closeness between the
The main latent independent variables were nights spent with grandchild (X1) and
the closeness between the grandparent and the grandchild. The X1 variable, number of
nights the grandparent spent with their grandchild without the parent of the child, was
only one indicator of the latent variable, nights spent with grandchild. X1 was measured
into a continuous variable that ranged from 0 to 365 days. Another latent independent
variable, the closeness between the grandparent and the grandchild, was measured
according to three indicators: (1) frequency of meeting with grandchild (X2), (2)
frequency of communicating with grandchild (X3), and (3) grandparent’s self rating of
53
closeness with grandchild (X4). X2 and X3 were measured into ordinal level data,
whereas X4 was a continuous variable that ranged from 0 (not at all close) to 10
(extremely close). The Cronbach’s alpha test of reliability among X2, X3, and X4
was .697.
observed variables – happiness (Y1), quality of life (Y2), depression (Y3), and
reliability among Y1, Y2, Y3, and Y4 was .589. Two instruments, happiness and quality
of life, were from the Institute for Social Research at University of Michigan. The
variable RT201 (Y1), measuring how happy the respondents were, ranged from 1 (very
unhappy) to 7 (very happy). It was measured through global positive affect measurement
from the quality of life survey from the Institute for Social Research at the University of
Michigan.
Eleven variables, RT202A~RT202K and quality of life measure (Y2), were also
from the quality of life surveys at the University of Michigan. The measurement scale
was ordinal, ranging from 1 (very dissatisfied) to 7 (very satisfied). While the
measurement originally consisted of 11 items, the researcher excluded one item (RT202K)
for this study. The excluded item was intended to measure how satisfied the respondents
were with their present job. However, the majority of the sample in this study did not
answer this question, whether they worked or not. Thus, it was difficult to identify the
54
missing cases on the RT202K because the respondents did not work at the time of data
collection, the respondents were reluctant to answer the question, or their answer was
randomly missing. Hence, the item RT202K, was excluded from this study. The
researcher conducted the internal consistency reliability test using Cronbach’s alpha; the
alpha value did not change much after excluding RT202K. It was .836 with all eleven
items and .827 after excluding RT202K. In this study, the total summated scores of the 10
items were used to measure the quality of life, Y2. Therefore, the higher the Y2 score, the
Depression, Y3, was measured through a modified version of the Center for
Epidemiologic Studies Depression Scale (CES-D). The measure asked how many days
during the past week the respondent felt depressed for 12 occasions. While the CES-D
was originally coded from 0 (0 days) to 7 (7 days a week), the researcher recoded it from
0 (depressed 7 days a week) to 7 (0 days). Thus, as the Y3 score gets higher, it means the
respondent is less depressed. The CES-D is applied to adults in general, compared to the
Geriatric Depression Scale (GDS), which is designed especially for older adults. Since
the age of this study’s population varies from mid-30s to late-90s, the CES-D may be
appropriate to interpret their level of depression. The original version of the CES-D has
retest reliability ranges from .51 to .67 in short intervals and .41 to .54 in longer intervals.
Concurrent validity is the highest with the Hamilton rating scale (.50s~.80s) and varies
from 0.30s to 0.80s with other depression scales (Burns, Lawlor, & Craig, 2004).
Beekman, Deeg, Van Limbeek, Braam, De Vries, and Van Tilburg (1997) examined the
55
criterion validity of the CES-D with an older population, between 55 and 85 years old.
When the cut-off score was set at 16, sensitivity was 100 % and specificity was 87.6 %.
As the cut-off score rose, the sensitivity tended to be lower, yet the specificity became
Ross and Mirowsky (1988) implemented the 12-item CES-D in their study, “Child care
and emotional adjustment to wives’ employment,” and the internal consistency was .85
for wives and .81 for husbands. The researcher also tested the internal consistency using
scales. The scale consisted of eighteen variables, RT217A through RT217H and RT218A
through RT218J. The original scale ranged from 1 (strongly agree) to 5 (strongly
agree) so that as the score gets higher, the respondent’s well-being level gets higher.
statements and 16 negative statements. Its rating scale was a six-point scale, ranging from
strongly agree to strongly disagree. The scale is divided into six sections – (1) self
acceptance, (2) positive relations with others, (3) autonomy, (4) environmental mastery,
(5) purpose in life, and (6) personal growth. The internal consistency reliability of the 20-
item scale varied from .86 to .93 depending upon each section. In addition, the test-retest
reliability of the 20-item scale over a 6-week period ranged from .81 to .88. To measure
validity, the researcher compared this scale with previous measures of both positive and
negative functioning, such as life satisfaction, self esteem, and depression. The results
56
indicated that all the positive functioning items were positively correlated with previous
positive measures, which ranged from .25 to .73. Similarly, correlations with the negative
functioning items were negative, and ranged from -.30 to -.60. The reliability of the 18-
item scale used in NSFH was evaluated again by Ryff in the 1990s. Ryff and Keyes
(1995) stated that the internal consistency of the 18-item scale ranged from .33 to .56.
While the reliability score that Ryff and Keyes yielded was not high, the researcher’s
own reliability score for this study using Cronbach’s alpha was .70.
In addition, three observed variables measured the latent variable, cultural attitude.
The three items measuring cultural attitude are from a set of family related attitude items
developed by a team at the University of Wisconsin. The variables asked whether the
respondent thought elderly parents should live with children (C1), whether the respondent
thought children should help older parents with money (C2), and whether the respondent
thought adult children with problems should live with a parent (C3). The respondents’
answers were coded into the five level ordinal data that ranged from strongly disagree to
strongly agree. The Cronbach’s alpha of the reliability among C1, C2, and C3 was .515.
Although the reliability increased to .582 if the C3 was deleted, the researcher decided to
include C3 because the Cronbach’s alpha would not increase enough to reach above .70.
57
The grandparent’s functional capacity was measured with the total score of
limitations on daily living. It was a set of items asking whether the respondent had
physical and mental limits for daily living. The items consisted of seven questionnaires
that were measured into three ordinal levels: (1) do not limit at all, (2) limit a little, and (3)
limit a lot. The researcher used the total score of the seven items so that the latent
variable, the grandparent’s functional capacity, has only one observed variable.
The part of the survey that addresses the grandparent-parent support consists of
six dichotomous questions asking whether the grandparent and the parent give and
receive support from each other during a month. Three questions addressed whether the
grandparent provided support to his or her adult child, and another three questions asked
whether the grandparent received support from his or her adult child. The researcher also
combined the six variables into one indicator to the latent variable, grandparent-parent
Statistical Analysis
latent independent variables, the nights spent with grandchild and the closeness between
the grandparent and the grandchild, were analyzed. Also, the range and the mean of the
58
Descriptive statistics of each observed variable for measuring cultural attitude,
addition, the number of grandchildren in the family, the age of both the grandparents and
the grandchild, and the grandparent’s gender, ethnicity, and the presence or absence of
legal custody were also measured in order to identify general characteristics of the study
sample.
correlations between the age of the grandparents, the age of the grandchildren, and the
grandparents and grandchildren, it was possible to figure out which grandparents are
To test the six hypotheses, Structural Equation Modeling (SEM) was used.
LISREL 8.8 was used for the SEM analysis, while descriptive statistics were analyzed
with SPSS 17.0. Model one was used to test hypothesis one. Hence, four independent
variables and two dependent variables were included in model one. The four independent
variables were (1) the number of grandchildren in family; (2) the grandparent’s age; (3)
the youngest grandchild’s age; and (4) the oldest grandparent’s age. The number of nights
spent with the grandchild and the closeness between the grandparent and the grandchild
59
Model two is related to hypothesis test two, which is intended to measure two
being. Thus, three latent variables, (1) nights spent with grandchild; (2) closeness
between the grandparent and the grandchild; and (3) grandparent’s well-being, were
Model three had three more variables added to the three variables included in the
model two. The researcher intended to measure the effect of cultural attitude and the
grandparent-parent support on the relationships among the number of nights spent with
the grandchild, the closeness between the grandparent and the grandchild, and the
in the model three in order to address its impact on the grandparent’s well-being.
Three multigroup comparison models were applied in this study to test hypotheses
four through six. The variables included in three multigroup models were the same as
those used in the model three. However, model four focused on gender difference: the
different coefficients for the path among the number of nights spent with the grandchild,
the closeness between the grandparent and the grandchild, and the grandparent’s well-
being. Model five tested the impact of ethnic differences in the model. The ethnicity
American grandparents. The last multigroup model comparison, model six, measured
paths among the number of nights spent with grandchild, the closeness between the
60
grandparent and the grandchild, and the grandparent’s well-being. As supplements to the
multigroup comparison, T-test results were also presented in order to address differences
61
Chapter 5: Research Findings
Initial Analyses
Sample demographics are shown in Table 5.1. through Table 5.8. below. As
described in Table 5.1., grandparents spent nights with their grandchildren for about three
weeks on average. However, the number of nights spent with the grandchild varied
depending upon the grandparents’ characteristics; thus, some grandparents actually lived
with their grandchild while others had not spent a night with their grandchildren for a
frequently; 33.1% of grandparents answered that they meet with their grandchild more
than once a week, followed by several times a year (20.9%), 1 to 3 times per month
(19.7%), and about once a week (17.2%). Approximately 25% of grandparents reported
communicating with their grandchildren more than once a week, and 21.5% of
grandparents answered that they communicated with their grandchild more than once a
month. Lastly, grandparents tended to think that they were very close to their
grandchildren: more than 50% of grandparents in this study thought that they were
extremely close to their grandchild, and more than 30% of the grandparent respondents
62
Frequency %
Nights spent with grandchild last year (X1)
Mean 22.05 days
Median 2 days
Minimum 0 day
Maximum 365 days
Frequency of meeting grandchild per year (X2)
Not at all 80 3.2
About once a year 138 5.5
Several times a year 520 20.9
1 to 3 times a month 490 19.7
About once a week 429 17.2
More than once a week 826 33.1
Grandchild lives with grandparent 10 0.4
Frequency of communication with grandchild per year (X3)
Not at all 196 8.8
About once a year 68 3.0
Several times a year 446 19.9
1 to 3 times a month 482 21.5
About once a week 468 20.9
More than once a week 572 25.6
Grandchild lives with grandparent 6 0.3
Closeness with grandchild (X4)
Not at all 30 1.2
1-5 231 9.3
6-9 879 35.3
Extremely close (10) 1350 54.2
Table 5.1. Closeness between the Grandparent and the Grandchild
Table 5.2. shows grandparents’ well-being scores. Happiness scores ranged from
1 to 7, where 7 is very happy. Hence, grandparents in the sample tended to feel that they
63
are happy about their overall life. Quality of life scores ranged from 7, very dissatisfied to
all 10 items, to 70, very satisfied to all 10 items. As the mean score of the quality of life
measure is 55.26, the respondents in this study were relatively satisfied their overall life.
Depression was measured with scores from 0 to 84. 0 means the respondent felt
depressed on every item seven days a week, whereas 84 means the respondent never felt
depression on any item. While depression score ranges were wider compared to other
well-being measures, the grandparents in the sample tended not to be depressed. Lastly,
psychological well-being scores range from 0 to 90, where 90 means the respondent’s
the overall psychological well-being of the grandparents was not low, either.
Grandparents’ cultural attitudes toward family related issues are shown in Table
5.3. Interestingly, more than half of the grandparents answered that elder parents should
live with their adult children. Also, the majority of the grandparents expected financial
64
support from their child as they got older. In addition, more grandparents thought adult
children with problems should live with parents. Grandparents in this study tended to
emphasize family interactions and open exchanges of support among family members.
Frequency %
Elder parents should live with (adult) kids (C1)
Strongly disagree 40 1.6
Disagree 621 25.0
Neither agree nor disagree 502 20.2
Agree 1,215 49.0
Strongly agree 102 4.1
Child should help older parent with money (C2)
Strongly disagree 16 0.6
Disagree 355 14.2
Neither agree nor disagree 358 14.3
Agree 1,641 65.7
Strongly agree 128 5.1
Adult kids with problems can live with parent (C3)
Strongly disagree 54 2.2
Disagree 727 29.1
Neither agree nor disagree 437 17.5
Agree 1,227 49.1
Strongly agree 56 2.2
Table 5.3. Cultural Attitude
Table 5.4. shows how much grandparents gave and took support from their adult
children during the last month. The percentage was 80.3% of the grandparents who
65
answered that they provided emotional support to their adult child. In addition,
grandparents tended to help their adult children with childcare. In other words, 30% of
the grandparents provided childcare support for their adult child while he or she worked.
Also, almost 40% of the grandparents provided childcare support while their adult child
was not working. The grandparents tended to provide more help than they received from
their adult child. For instance, less than 30% of the grandparents received help with
shopping or transportation from their adult child during a month. Also, more than 70% of
the grandparents tended to do their house work by themselves without help from their
child.
(Yes) Frequency %
Grandparent gave emotional support to adult child
(GPPS1) 2,033 80.3
66
Grandparents did, however, tend to receive more emotional support from their
adult child. In addition, it may be important to point that it may not be the same
population who provided child care support to their adult child with those who received
in the study did not have physical and/or mental limitations to perform daily living skills.
On the other hand, it is assumed that due to the age, more than 40% of grandparents felt
Table 5.6. shows how many grandchildren grandparents had in their family.
Grandparents in this study had average five grandchildren in their family. While most
67
Mean 5.06
Median 4
Minimum 1
Maximum 30
Frequency %
1~9 2,228 87.7
10~19 270 10.6
20~29 30 1.2
30 or more 13 0.5
The ages of the grandparent and the grandchild are shown in Table 5.7.
Grandparents’ age varied from the 30’s to the 90’s. Grandparents in their 50’s were the
majority population in this study. Then, grandparents between 60 and 69 years old
followed, with 27%. While most of the grandparents were 50 years old and over, there
Compared to the grandparent’s age, the ages of the grandchildren were limited to
the youngest and the oldest grandchildren. While the majority of the grandparents
answered their youngest grandchild was under 5 years old, more than 6% of the
68
Frequency %
Age of the grandparent
Under 39 15 .6
40~49 281 11.1
50~59 814 32.1
60~69 685 27.0
70~79 453 17.9
80~89 258 10.2
90 or older 31 1.2
Age of the youngest grandchild
Under 1 year old 503 20.0
1~5 1159 46.0
6~12 493 19.6
13~19 204 8.1
20~29 117 4.6
30~39 40 1.6
40 or older 5 .2
Age of the oldest grandchild
Under 1 year old 82 3.3
1~5 460 18.2
6~12 712 28.2
13~19 498 19.7
20~29 505 20.0
30~39 212 8.4
40 or older 54 2.1
Table 5.7. Age of Grandparents and Grandchildren
grandparent’s ethnicity, and legal custody of their grandchild, were included. Gender of
69
the grandparents was drawn from the roster one data of wave three. As shown in Table
5.8., there were more grandmothers than grandfathers in the sample. The data of
grandparents’ ethnicity were from wave one; therefore, sample size was down to 1,660
after the researcher excluded missing values. Among 1,660 grandparents, more than 80%
respondents had ever had primary responsibility for their grandchild at the time of data
collection.
Frequency %
Gender
Male 833 32.8
Female 1,708 67.2
Ethnicity of grandparents
Caucasian 1,351 81.4
Non Caucasian 309 18.6
African American 239 14.4
Hispanic 62 3.7
American Indian 5 .3
Asian 3 .2
Ever had primary responsibility for grandchild
Yes 366 14.4
No 2,171 85.6
Table 5.8. Demographics of Grandparents
70
Hypotheses Testing with Structural Equation Modeling
The researcher assessed the model fit between the hypothesized model and the
sample data for all models included in this study. The model fit was measured through
Standardized Root Mean Square Residual (SRMR), Goodness of Fit Index (GFI), and
Root Mean Square Error of Approximation (RMSEA) because chi square statistics are
influenced by large sample size. The SRMR values below .05, GFI values above .90, and
RMSEA values below .05 are often considered as an acceptable fit (Schumacker &
Lomax, 2004). Most of the models tested in this study were within these criteria. To
examine relations among latent variables and between latent variables and observed
variables, the standardized path coefficients of each model were also assessed.
Statistically significant paths (p<.05) are indicated as solid lines, whereas nonsignificant
Throughout the models tested in this study, two measures of the grandparent-
being. In other words, the path from the nights spent with grandchild to the grandparent’s
well-being was negative. However, the closeness between the grandparent and the
the closeness between the grandparent and the grandchild may affect grandparents
Grandparents may feel positively about their close relationship with their grandchild. As
a result, the grandparents’ well-being tends to increase when they feel that they have a
close relationship with their grandchild. However, when grandparents frequently stay
71
overnight with their grandchild without the parent of the child, grandparents seem to feel
grandparents’ well-being negatively across all the models. For instance, those who
actively gave and took support with their adult child had lower well-being scores than
those who rarely exchanged support with their adult child. In addition, the closeness
between the grandparent and the grandchild had a positive impact on the cultural attitude
throughout every model. In contrast, the relationship between the nights spent with the
grandchild and cultural attitude was either positive or negative depending upon the
sample characteristics.
Hypothesis Test 1
Hypothesis one was to test if the age of both grandparents and grandchildren and
the number of grandchildren in the family affect the number of nights spent with
grandchild and the closeness between the grandparent and the grandchild. To test this
hypothesis, a multiple indicators and multiple causes (MIMIC) model with two latent
variables and four observed variables was applied to the model one. Total cases included
in the model were 1,822 after excluding missing values on each variable. In Figure 5.1.,
‘No.GC’ indicates the number of grandchild the grandparent had in family. ‘GP Age’
refers to the grandparent’s age. ‘GCY’ means the age of the youngest grandchild and
‘GCO’ means the age of the oldest grandchild. The model fit was within the acceptable
72
range. Chi square was 20.55 with a degree of freedom of 8, which resulted in a p-value of
Standardized
Paths
Estimates
Number of grandchild in family (No.GC) ->Nights spent with GC .09
Age of the grandparent (GP Age) -> Nights spent with GC -.15
Age of the youngest grandchild (GCY) -> Nights spent with GC .03 (NS)
Age of the oldest grandchild (GCO) -> Nights spent with GC -.03 (NS)
Number of grandchild in family -> Closeness between GP & GC .12
Age of the grandparent -> Closeness between GP & GC -.17
Age of the youngest grandchild -> Closeness between GP & GC -.08
Age of the oldest grandchild -> Closeness between GP & GC .03 (NS)
Goodness of Fit Indexes
SRMR .014
GFI 1.000
RMSEA .029
Chi square (df=8) 20.55
Note. NS: Not Significant (p=.05)
Table 5.9. Structural Model Result
In terms of path coefficients, all paths were significant, except for the three paths
shown in Table 5.9. The number of grandchildren the grandparent had in the family and
the grandparent’s age were significantly related to the number of nights spent with the
grandchild. That is, the more grandchildren a respondent had, the more likely that
grandparent was to stay overnight with those grandchildren. In addition, the older the
grandparent, the less likely they were to stay overnight with their grandchild. The number
73
of grandchildren the grandparent had in the family, the grandparent’s age, and the age of
the youngest grandchild were statistically related to the closeness between the
grandparent and the grandchild. In other words, the more grandchildren a respondent had,
the more likely the grandparent was to be close with their grandchild. Moreover, when
both grandparents and grandchildren were younger, they were more likely to be close to
each other than older grandparents with their grandchildren. The age of the oldest
grandchild was not significantly related to either nights spent with the grandchild or
74
75
Hypothesis Test 2
Hypothesis two was to determine whether the number of nights spent with
grandchild and the closeness between the grandparent and the grandchild are associated
with the grandparents’ well-being. To test hypothesis two, three latent variables and eight
indicators were included in the model two. One of the latent independent variables was
the nights spent with grandchild. Another independent variable was the closeness
between the grandparent and the grandchild. The dependent latent variable in the model
was the grandparents’ well-being. The sample size used in the model was 1,434 after the
researcher excluded missing values on each indicator. The model fit was within the
acceptable range, except for RMSEA. SRMR value was .037, GFI value was .99, and
RMSEA value was .054. Chi square value was 81.79 with a degree of freedom of 16;
Standardized estimates of path coefficients are shown in Figure 5.2. All paths
were significant at the alpha level of .05. The more frequently grandparents stayed
overnight with their grandchild, the lower their well-being was. On the other hand, the
closeness between the grandparent and the grandchild was positively related to the
grandparents’ well-being.
76
77
Hypothesis Test 3
support, and cultural attitude affect the grandparent-grandchild relationship and the
grandparents’ well-being.”
To test hypothesis three, six latent variables were included in the model. As in
model two, the independent variables were the number of nights spent with the
grandchild and the closeness between the grandparent and the grandchild. The dependent
variable, grandparents’ well-being, also remained the same as in model two. Two latent
variables, cultural attitude and grandparent-parent support, were included in the model
three to test mediating effects. Lastly, the grandparent’s functional capacity was included
as a control variable to the grandparents’ well-being. The total sample size in this model
was 1,396. All of the path coefficients shown in Table 5.10. were significant at the alpha
level of .05. The model fit was within an acceptable range, as is shown in Table 5.10.
However, the chi square value was significant, possibly due to the large sample size.
78
Factor Loadings Standardized Estimates
Nights spent with grandchild
X1 1.00
Closeness between GP & GC
X2 .76
X3 .64
X4 .64
Grandparent’s well-being
Y1 .65
Y2 .93
Y3 .77
Y4 .55
Cultural attitude
C1 .76
C2 .56
C3 .29
Grandparent-parent support
GPPS 1.00
Grandparent’s functional capacity
FC 1.00
Goodness of Fit Indexes
SRMR .040
GFI .980
RMSEA .044
Chi square (df=56) 205.51
Table 5.10. Measurement Model Result (H3)
Figure 5.3. indicates that the number of nights spent with the grandchild was not
to model two, cultural attitude and grandparent-parent support decreased the negative
relationship between the number of nights spent with the grandchild and the grandparents’
well-being. In addition, the two mediating factors increased the positive relationship
between the closeness between the grandparent and the grandchild and the grandparents’
79
well-being. Although the closeness between the grandparent and the grandchild was
illustrated between the closeness between the grandparent and the grandchild and cultural
attitude; however, the relation between cultural attitude and the grandparents’ well-being
was negative. The same patterns appeared in the relationship between the number of
grandparents’ well-being, although the paths from the number of nights spent with the
significant.
80
81
Hypothesis Test 4
according to whether the grandparent is male or female. The same six latent variables
were included in the test of the fourth hypothesis as were in hypothesis test three.
However, grandparents’ gender variable was included in this test to address gender
differences in the model. The total sample size in this model was 1,396: 870
grandmothers and 526 grandfathers. All the path coefficients shown below in Table 5.11.
were significant at the alpha level of .05. Also, both the grandmothers’ model fit and the
grandfathers’ model fit were within an acceptable range, as is shown in Table 5.11.
However, similarly to the model three, the chi square value was significant.
As described in Figure 5.4., in the grandmothers’ model, paths from the number
of nights spent with grandchild to cultural attitude and the grandparent-parent support
were not statistically significant. The grandfathers’ SEM model shown in Figure 5.5. had
four paths that were not statistically significant at the alpha level of .05. Hence, cultural
attitude did not affect the relationship between the closeness between the grandparent and
the grandchild or the grandparents’ well-being among grandfathers, while it did influence
nights spent with the grandchild. In addition, the impact of the closeness between the
grandparent and the grandchild on the grandparents’ well-being was lower among
82
being. The grandparent-parent support had a negative influence on the well-being of both
Standardized Estimates
Factor Loadings
Grandmothers Grandfathers
Nights spent with grandchild
X1 1.00 1.00
Closeness between GP & GC
X2 .76 .74
X3 .61 .66
X4 .60 .67
Grandparent’s well-being
Y1 .67 .60
Y2 .92 .98
Y3 .79 .68
Y4 .58 .48
Cultural attitude
C1 .79 .75
C2 .63 .45
C3 .27 .33
Grandparent-parent support
GPPS 1.00 .27
Grandparent’s functional capacity
FC 1.00 1.00
Goodness of Fit Indexes
SRMR .043 .041
GFI .980 .970
RMSEA .042 .034
Chi square (df=56) 141.23 90.82
Table 5.11. Measurement Model Result (H4)
83
84
85
The T-test results for testing differences of well-being scores between
>.05; the quality of life (Y2) was t (1,889)=1.566, p >.05; depression (Y3) was t
Depression was the only category that was statistically significant across gender,
Hypothesis Test 5
Hypothesis test five also had used the same six latent variables as hypothesis tests three
and four. In order to identify ethnic differences, 717 Caucasian American grandparents
and 165 non-Caucasian American grandparents were included in the models. The total
sample size was decreased from 1,396 to 882 due to the missing values on the ethnicity
All the path coefficients shown below in Table 5.12. were significant at the alpha
grandparents had the highest factor loading on frequency of meeting, X2, whereas non-
86
of life, Y2, had the highest factor loading among Caucasian American grandparents,
while depression, Y3, had the highest factor loading among non-Caucasian American
grandparents.
Standardized Estimates
Factor Loadings
Caucasian GP Non-Caucasian GP
Nights spent with grandchild
X1 1.00 1.00
Closeness between GP & GC
X2 .76 .67
X3 .64 .74
X4 .69 .52
Grandparent’s well-being
Y1 .68 .35
Y2 .95 .53
Y3 .80 .87
Y4 .57 .57
Cultural attitude
C1 .78 1.00
C2 .52 .25
C3 .32 .24
Grandparent-parent support
GPPS 1.00 1.00
Grandparent’s functional capacity
FC 1.00 1.00
Goodness of Fit Indexes
SRMR .048 .056
GFI .970 .940
RMSEA .045 .024
135.99 63.68
Chi square
(df=56) (df=58)
Table 5.12. Measurement Model Result (H5)
87
The model fits for both the Caucasian American and the non-Caucasian American
grandparents’ models that are described in Table 5.12. Although GFI and RMSEA values
for both models were in an acceptable range, the SRMR value of the non-Caucasian
American grandparents’ models was higher than .05. The SRMR value did not decrease
to the acceptable level after model modifications. On the other hand, the chi square value
of the non-Caucasian American grandparents’ model was not significant (p=.28). The
degrees of freedom of each model differed because the Caucasian American grandparents’
model had two more measurement error covariances than the non-Caucasian American
grandparents’ model.
Caucasian American grandparents are shown above in Figure 5.6. and Figure 5.7. The
Caucasian American grandparents’ model had significant paths between the nights spent
with the grandchild and the grandparents’ well-being, and between the closeness between
the grandparent and the grandchild and the grandparents’ well-being. On the other hand,
two paths were not significant in the non-Caucasian American grandparents’ model. The
Caucasian American grandparents’ model demonstrated the same patterns among the
number of nights spent with the grandchild, the closeness between the grandparent and
the grandchild, and the grandparents’ well-being as those in the model three. In other
words, their well-being tended to decrease as they spent more nights with their grandchild.
However, the closeness between the grandparent and the grandchild positively influenced
positive relationship with the closeness between the grandparent and the grandchild,
88
while the grandparents’ well-being tended to decrease as they exchanged support and
help more frequently with their adult child. The closeness between the grandparent and
the grandchild was positively related to the cultural attitude as well, although cultural
attitude was not statistically significantly related to the grandparents’ well-being. Cultural
attitude did not affect the relationship between the number of nights spent with the
American grandparents’ model was limited in its ability to explain the relationships
among the six latent variables, possibly due to the small sample size. Although most
more from a functional capacity regarding their level of well-being. In addition, the
closeness between the grandparent and the grandchild was less likely to relate to the
The T-test results are as follows: happiness (Y1) was t (1,219)=-0.561, p >.05; the
quality of life (Y2) was t (1,219)=2.294, p=.022; depression (Y3) was t (1,219)=5.674,
although the results were not significant. In contrast, Caucasian American grandparents
scored higher on quality of life and level of depression, and the results were statistically
significant.
89
90
91
Hypothesis Test 6
was related to the legal custody of grandparents. The same six latent variables were
included in models three, four, and five. The total sample size in hypothesis test six was
1,396. Among 1,396 grandparents, noncustodial grandparents were 1,198 and custodial
The standardized path coefficient results and the model fit were shown above in
Table 5.13. Except for the path of the three cultural attitude variables in the custodial
grandparents’ model, all other paths were statistically significant at the alpha level of .05.
Similar patterns of factor loading were shown in the noncustodial grandparents’ model
and the custodial grandparent’s model. In other words, frequency of meeting with
grandchild, X2, was the highest factor loading in both noncustodial and custodial models.
However, Y2, quality of life, had the highest factor loadings in the noncustodial
grandparents’ model while Y3, depression, was the highest in the custodial grandparents’
model. The model fit of noncustodial grandparents was within an acceptable range;
however, the p-value of the chi square was below .05. The model of the custodial
grandparents had .067 for its SRMR and a significant chi square value, while GFI and
92
Standardized Estimates
Factor Loadings
Custodial GP Noncustodial GP
Nights spent with grandchild
X1 1.00 1.00
Closeness between GP & GC
X2 .78 .75
X3 .63 .63
X4 .46 .66
Grandparent’s well-being
Y1 .56 .64
Y2 .59 .71
Y3 .84 .69
Y4 .52 .59
Cultural attitude
C1 .64 (NS) .77
C2 .76 (NS) .55
C3 .15 (NS) .32
Grandparent-parent support
GPPS 1.00 1.00
Grandparent’s functional capacity
FC 1.00 1.00
Goodness of Fit Indexes
SRMR .067 .040
GFI .940 .970
RMSEA .050 .046
Chi square (df=57) 91.98 201.49
Note. NS: Not Significant (p=.05)
Table 5.13. Measurement Model Result (H6)
is shown in Figure 5.8. and Figure 5.9. It was consistent in the noncustodial grandparents’
models that the path from the number of nights spent with the grandchild to the
grandparents’ well-being was negative, whereas the path from the closeness between the
grandparent and the grandchild to the grandparents’ well-being was positive. However, in
93
the custodial grandparents’ model, both paths had negative impacts on the grandparents’
well-being, although these paths were not significant. The path from the closeness
between the grandparent and the grandchild to the grandparents’ well-being was
The T-test results are as follows: happiness (Y1) was t (1,887)=-2.743, p=.006;
quality of life (Y2) was t (1,887)=-4.265, p<.001; depression (Y3) was t (1,887)=-4.922,
it is assumed that the well-being of non-custodial grandparents was higher than custodial
grandparents.
94
95
96
Chapter 6: Discussion
In this study, there are limitations regarding variables not included in the analysis.
earlier in the method section, the dataset used in this study does not include each
grandchild’s age. Thus, the researcher used the age of the oldest grandchild and the age
of the youngest grandchild to determine the influence of age on the closeness between the
grandparent and the grandchild. The theoretical framework of this study is based on a life
course perspective and Bowen family systems theory. The life course perspective
emphasizes changes in individuals’ roles in the family as they age. In addition, the life
course perspective addresses how individual time, generational time, and historical time
influence individuals’ family life. In this respect, the age of the grandchild may be a
crucial factor in analyzing the relationship between the grandparent and the grandchild.
Although the age of the youngest grandchild may be related to the age of the oldest
grandchild and the age of the grandparent, a more accurate analysis would result if the
Another limitation is that there were no variables available for analyzing the
relationship between the grandchild and the parent of the child. As this study’s theoretical
framework is based on Bowen family systems theory, the main purpose of the study was
to examine whether the triangular relationship among the grandparent, parent, and
97
grandchild influences the well-being of the grandparent. Although the study’s focus is on
the relationship between the grandparent and the grandchild, two other aspects (parent-
comprise the triangular relationship between the grandparent, parent, and grandchild.
relationship and the grandparent-parent support. Therefore, it was not possible to address
relationship and the parent-grandchild relationship in the analysis. In other words, Bowen
family systems theory assumes that the triangular relationship decreases the anxiety of
the relationship between any two of the constituent elements (grandparent, parent, and
grandparent and the grandchild may be closely related to the relationship between the
grandparent and the parent. For instance, the parent whose relationship with his or her
older parent is not close may be less likely to bring his or her child to that older
grandparent than those whose relationship with their older parent is good. The status of
the triangular relationship among the grandparent, parent, and grandchild is dependent
upon whether the status of each relationship is positive or negative. On the other hand, in
98
this study, the researcher was not able to assess whether each relationship was positive or
negative. Hence, it was not possible to ascertain why some grandparents rarely give
support to and receive support from their adult child, while others do frequently.
this study. In other words, the frequency of communication and/or meetings with the
grandchild may be closely related to the physical distance between the grandparent and
the grandchild. It is easily assumed that the grandparent who lives next to the grandchild
may have more chances to meet his or her grandchild than those who live in other states.
While physical distance often influences emotional distance between people, the
variables examining such aspects were not available in the dataset used in this study.
relationship between grandparents who live close to their grandchild and those who live
farther away.
The lack of ethnic diversity in the subjects of this study is also one of its
American. Among 309 non-Caucasian American grandparents, 239 (77%) were African
American grandparents. However, there were only five American Indian grandparents
and three Asian American grandparents. Hence, the comparison between Caucasian
non-Caucasian American grandparents included did not have a large enough non-African
99
Lastly, the dataset is missing many values for variables that asked about economic
status. Previous studies have shown that grandparents’ well-being depends on their
economic status. The dataset has variables such as hours worked per week, gross income,
whether the respondents worked full-time or part-time and so on. Although these
variables included a sufficient number of cases, after filtering out those cases that
included at least one grandchild in the family, the number of cases became too small to
Six hypotheses were tested in this study. The first hypothesis was to address the
relationship between grandparents’ and grandchildren’s life courses and their closeness to
each other. While the grandparent’s age affected two closeness latent variables (the
number of nights spent with grandchild and the closeness between the grandparent and
the grandchild), the age of the grandchildren was not significant to the two direct
significant to the closeness variables. Two indicators out of four significantly affected the
two closeness latent variables, the number of nights spent with grandchild and the
closeness between the grandparent and the grandchild. The age of the youngest
grandchild was significantly, but inversely related to the closeness between the
grandparent and the grandchild; however, it did not affect the number of nights spent
with the grandchild. Thus, the grandchild is less likely to be close with the grandparent as
100
Although the life course perspective assumes that individuals tend to have a
different family interaction based on their life course, as shown in Figure 6.1., this study
only confirmed that the grandparents’ life course influenced the closeness with their
grandchildren. There are three life courses – the grandparent’s life course, the parent’s
life course, and the grandchild’s life course. Each individual’s life course, which inter-
correlates with every other family member’s life course, makes a whole family life course
that is unique to each family. For instance, the life course of the 19-year-old parent who
has a 2-year-old child may be totally different from that of 35-year-old parent who also
has a 2-year-old child. These two different parent life courses would result in different
life courses for their children, even though the two children are in the same generation. In
the same way, life courses of the grandparents of these two children would be different.
interact with each other; however, in this study, grandparents’ life courses and their effect
was highly correlated with the age of the oldest grandchild, the oldest grandchild’s age
was not statistically related to the two closeness latent variables. Also, younger
grandparents tended to have a closer relationship with their grandchild compared to older
grandparents. Thus, the first hypothesis was partially confirmed by this study.
101
Individual time
Generational time
Age of
grandparent
Number of
grandchildren
Figure 6.1. Key Concepts of Life Course Perspective and Variables in This Study
The second hypothesis was fully confirmed by this study, in that the number of
nights spent with the grandchild and the closeness between the grandparent and the
indicators of the three latent variables were also statistically significant. Thus,
grandparents who spent more nights with their grandchild tended to have lower well-
being compared to those who spent fewer nights with their grandchild. On the other hand,
grandparents who had more frequent communication and meetings with their grandchild
102
tended to have higher well-being compared to those who had less contact with their
grandchild.
relationship and the grandparent’s well-being. This hypothesis was confirmed by the
attitude indirectly increased the positive score of the closeness between the grandparent
and the grandchild. Similarly, these factors indirectly decreased the negative score
between number of nights spent with the grandchild and the grandparent’s well-being.
However, direct effects of cultural attitude and the grandparent-parent support on the
it was possible to address whether the triangular relationship in Bowen family systems
theory might be applied to this study, as shown in Figure 6.2. Bowen family systems
theory assumes that when a third person is added to an anxious relationship between two
people, the stress of that third person increases as the anxiety between two people
decreases. The result of this study partially supports the first assumption – the stress of
the third person (grandparents) increases. After testing hypothesis three, it was confirmed
that as grandparents and parents exchanged support more frequently, the grandparent’s
well-being decreased. In addition, as grandparents spent more nights with the grandchild,
the grandparents’ well-being decreased. On the other hand, a higher score for the
closeness between the grandparent and the grandchild, measuring the frequency of
103
communication and meetings between the grandparent and the grandchild, resulted in a
higher score for the grandparent’s well-being. This pattern was consistent when testing
hypotheses four, five, and six. This means that the grandparent may not feel stressed until
higher level of the grandparent’s well-being. However, it was not possible to address
whether the anxiety between the parent and the grandchild decreases as the grandparent
spends more nights and/or communicates more frequently with the grandchild. This kind
of analysis would be feasible if there were variables measuring the anxiety between
Grandchild Parent
DV: Grandparent’s
well-being
Figure 6.2. Key Concepts of Bowen Family Systems Theory and Variables in This Study
104
The fourth hypothesis was to examine the difference between grandmothers and
grandfathers. As described in the results section, patterns of the paths among the
grandparents’ well-being, the number of nights spent with the grandchild, and the
closeness between the grandparent and the grandchild were similar between
grandmothers and grandfathers. On the other hand, the invariance test using chi square
characteristics; thus, it was impossible to further analyze differences in the variances and
American grandparents were not invariant, the patterns of the major paths were similar.
On the other hand, the SEM model used in this study was limited in its ability to explain
results for non-Caucasian American grandparents. The majority of the paths were not
partially confirmed through the SEM analysis. The variable, the nights spent with
grandchild, was less negatively correlated with the grandparent’s well-being among
noncustodial grandparents. Similarly, the closeness between the grandparent and the
105
custodial grandparents. Even though the relations among these variables were not
significant in the custodial grandparents’ model, it was assumed that staying overnight
with grandchildren was more burdensome for custodial grandparents because of their
legal responsibility for their grandchild. In addition, the impact of frequency of meeting
custodial and noncustodial grandparents; the impact was primarily among the
this obligation, but not necessarily by frequency of interaction, which was more
important in the model testing noncustodial grandparents. On the other hand, the amount
Most paths were not significant and the model fit among the custodial grandparents’
This study yielded both similarities with and differences from previous research.
First, this study was consistent with previous research (e.g. Lin & Harwood, 2003;
Lemme, 2002; Strom & Strom, 1997) regarding the grandparent’s age and the
relationship between the grandparent and the grandchild. That is, it was confirmed that
the older the grandparents, the less contact they have with their grandchildren. Older
grandparents were less likely to spend nights with their grandchild than were younger
grandparents differed in well-being. Based on the results from the t-test results, custodial
106
grandparents demonstrated lower well-being than noncustodial grandparents. In addition,
path significances among the number of nights spent with the grandchild, the closeness
between the grandparent and the grandchild, and the grandparent’s well-being also
the custodial grandparents’ SEM model had less significant paths than the noncustodial
grandparents’ model, both spending nights with the grandchild and the closeness between
the grandparent and the grandchild tended to be burdensome for custodial grandparents.
In addition, this study’s results were consistent with the previous research (McGowen,
Ladd, & Strom, 2006) that indicates that spending nights with the grandchild is stressful
whether the grandparents are custodial or noncustodial. McGowen, Ladd, and Strom
found that grandparents’ life satisfaction was lower when grandchildren were living with
them than among grandparents who did not have grandchildren living with them.
The major uniqueness of this study is its analysis of the closeness between the
grandparent and the grandchild and its effect on the grandparent’s well-being. Previous
research has focused mainly grandparents’ well-being comparing them in several respects,
the interactions between and among grandparents, parents, and grandchildren. However,
in this study, the researcher combined these two strands of research. The study’s focus
was on identifying the relationships between the grandparent’s well-being and the family
107
interactions using three latent variables – the number of nights spent with grandchild, the
closeness between the grandparent and the grandchild, and the grandparent-parent
support.
In addition, this study applied SEM analysis, although regression techniques have
been popular methods used to analyze these relationships. In previous studies, logistic
regression and t-test were applied to compare differences among two or more groups. For
instance, the researchers compared two or more populations, such as grandfathers and
and/or custodial and noncustodial grandparents (e.g. Kolomer & McCallion, 2005;
Reitzes & Mutran, 2004; Vandell, McCartney, Owen, Booth, & Clarke-Stewart, 2003). In
this study, however, the researcher included the factors related to the well-being of the
grandparents in the SEM analysis in order to explain the relationships among these
multiple factors. Thus, it was possible to test diverse models to find the best fit to explain
among independent variables and control variables as well as the relationship between
For instance, they sometimes become a third person in the parent-grandchild relationship
in order to reduce anxiety between the adult child and the grandchild. Some grandparents
also enter into the parent-grandchild relation just as a normal way of family interaction.
108
Thus, a grandparents’ level of stress may be depending upon the fact whether they are
with their grandchild by choice or not and how much time and energy they devote to
spending nights with the grandchild without the parent of the child worsened
grandparents’ well-being in this study. Goodfellow and Laverty (2003) also stated that
taking care of grandchildren was a physically and emotionally exhausting job for both
significant amounts of time for grandparenting, whether the grandparents are custodial or
counseling, and financial support programs are available for custodial grandparents in
many states. On the other hand, noncustodial grandparent caregivers who feel just as
stressed as custodial grandparents rarely receive formal support. The state of Georgia
provides child care subsidy as financial incentives to eligible grandparents who are over
60 years old. Cuyahoga County in Ohio provides cash assistance and day care vouchers
for low-income grandparents if they meet the eligibility guidelines. However, support for
Therefore, policy advocacy is necessary for more noncustodial grandparents who are
among available services for grandparent caregivers, counseling, tutoring, respite services,
legal information, stress reduction programs, and information and referral services are the
109
most needed services. Especially, as a specialized service for grandparent caregivers,
family relationships. They may newly define their relationship with their adult child and
grandchild. Custodial grandparents may need to accept their role as a legal guardian to
their grandchild. There may be resistance from the grandchild and/or from themselves
their grandchild may need to re-define their role in family. There is no clear definition for
noncustodial grandparents of their tasks. Their tasks may be changeable according to the
schedule of their adult child. Hence, the ambiguity of the noncustodial grandparents’
roles may result in family conflict, especially between the grandparent and the adult child.
Thus, Bowen’s family intervention may be helpful for grandparent caregivers to reduce
person who feels stress after involved in the triangle family relationship. Hence,
grandparents raising grandchildren may get benefits through Bowen family counseling as
to seek stress reduction strategies and set their roles in the family.
support for grandparents raising their grandchildren. Although the AARP website offers
often go to a local service center to join a grandparent support group or receive training.
On the other hand, those who take care of young children may not join programs that are
organized according to a fixed timetable. Additionally, some grandparents may have free
time in the evening. Hence, linking off-line resources with the on-line community may be
110
beneficial for those who need to seek assistance at any time. Moreover, the on-line
community may be able for grandparent caregivers to make a social network with their
grandchildren is cultural diversity. As indicated previously, there are various factors that
affect the well-being of the grandparents. Each culture, each generation, and each family
may have different views about family ties; thus, social work institutions may continue to
educate students about cultural diversity and respect cultural differences. In addition,
social workers should be culturally aware when working with grandparents taking care of
their grandchildren.
For future studies of this topic, it will be necessary to include factors not included
in this study, such as the grandchild’s age, family conflict, the grandparent’s physical
distance from the grandchild, and the grandparent’s economic status, to identify the
and the well-being of the grandparent. In addition, it may be beneficial if future studies
are able to include more non-Caucasian American grandparents in the sample. In this
study, the ratio of Caucasian to non-Caucasian American grandparents was about 8:2.
the same, it was regrettable that this study only included five American Indian and three
111
Different types of SEM models may be necessary to address ethnic variations and
differences related to the legal custody of the grandparents. The SEM model used in this
study did not explain the characteristics of non-Caucasian American grandparents and
The non-Caucasian American grandparents’ SEM model and the custodial grandparents’
SEM model contained many non-significant paths. Also, the model fit of both models
was lower than other models tested in this research. Thus, it is assumed that the two
populations may have other factors, not included in this research, that affected
of grandchildren. These research findings indicated that the relationship between the
nights spent with the grandchild and the grandparent’s well-being decreased after adding
relationship and the grandparent’s well-being strengthened after adding the above three
and strengthened the positive relationship. On the other hand, in future research, it may
be necessary to find out what characteristics of each variable affected the result that the
negative relationship was weakened and the positive relationship was strengthened.
background; for instance, whether they are from a culture that emphasizes individual
112
freedom or family ties. The model used in this study poorly explains non-Caucasian
grandparents. This may due to the complex factors associated with non-Caucasian culture.
Among Asian cultures, for instance, the nature of Chinese culture and Japanese culture is
totally different. Hispanic culture, which is comprised of diverse ethnic groups, also
differs from many Caucasian American norms relating to grandparenting. Thus, social
workers must carefully assess these cultural differences and tailor services based on these
ethnic variations.
generational time, an important construct underlying the life course perspective, affects
the triangular relationship among three generations. In this study, the ages of both
grandparents and grandchildren were included to identify its effects on the closeness
between the grandparent and the grandchild and the number of nights spent with the
grandchild. The age variable was not, however, included for testing hypotheses three to
six. Since the age of both grandparents and grandchildren may affect the relationship
113
References
AARP. (2006a). Grandparenting. Retrieved October 14, 2006 from the World Wide Web:
http://www.aarp.org/families/grandparents/.
AARP. (2006b). The grandparent study 2002 report. Retrieved September 30, 2006 from
http://www.aarp.org/research/family/grandparenting/aresearch-import-488.html
Ackerman, N. J. (1984). A theory of family systems. New York: Gardner Press, Inc.
Anderson, R. E., Carter, I. & Lowe, G. R. (1999). Human behavior in the social
De Gruyter.
Ashford, J. B., LeCroy, C.W., & Lortie, K. L. (2001). Human behavior in the social
Brooks/Cole.
245.
Becvar, D. S., & Becvar, R. J. (2006). Family therapy: A systematic integration (6th ed.).
Becvar, D. S., & Becvar, R. J. (2009). Family therapy: A systematic integration (7th ed.).
114
Beekman, A. T. F., Deeg, D. J. H., Van Limbeek, J., Braam, A. W., De Vries, M. Z., &
Van Tilburg, W. (1997). Criterion validity of the center for epidemiologic studies
Biddle, B. J. (1979). Role theory: Expectations, identities, and behaviors. New York:
Academic Press.
Burns, A. S., Lawlor, B. A., & Craig, S. (2004). Assessment scales in old age psychiatry.
Clausen, J. A. (1986). The life course: A sociological perspective. Englewood Cliffs, NJ:
Prentice Hall.
Collier, P. J., & Callero, P. J. (2005). Role theory and social cognition: Learning to think
Crouter, A. C., McHale, S. M., & Bartko, W. T. (1993). Gender as an organizing feature
Publishing Company.
115
implications for models of cultural transmission. Journal of Cross-Cultural
Fingerman, K. L., & Bermann, E. (2000). Applications of family systems theory to the
George, L. K., & Gold, D. T. (1991). Life course perspectives on intergenerational and
Gladding, S. T. (2002). Family therapy: History, theory, and practice (3rd ed.). Upper
Goodfellow, J., & Laverty, J. (2003). Grandparents supporting working families. Family
Hagestad, G.O. (2003). Interdependent lives and relationships in changing times: A life
course view of families and aging. In R.A. Settersten Jr. (Eds.), Invitation to the
life course: Toward new understandings of later life (pp. 135-160). Amityville,
116
Hahlweg, K. (2004). Strengthening partnerships and families. In P. L. Chase-Lansdale, K.
University Press.
Hareven, T. K. (1994). Aging and generational relations: A historical and life course
Hareven, T. K. (2000). Families, history, and social change: Life course and cross-
Hayslip Jr. B., Henderson, C. E., & Shore, R. J. (2003). The structure of grandparental
Hayslip Jr., B., & Kaminski, P. L. (2005). Grandparents raising their grandchildren: A
review of the literature and suggestions for practice. The Gerontologist, 45(2),
262-269.
Ikeda, K., & Hatta, T. (2001). Perceptions of family structures by Japanese students. In
T.E. Gehring, M. Debry, & P.K. Smith, P. K (Eds.), The family system test FAST:
117
Julian, T.W., McKenry, P.C., & McKelvey, M.W. (1994). Cultural variations in
Kleiner, H .S., Hertzog, J., & Targ, D. B. (1998). Grandparents acting as parents.
http://www.uwex.edu/ces/gprg/article.html
Lemme, B. H. (2002). Development in adulthood (3rd ed.). Boston: Allyn and Bacon.
Maccoby, E. E. (2003). The gender of child and parent as factors in family dynamics. In
118
neglected side of family relationships (pp. 191-206). Mahwah, NJ: Lawrence
Macmillan, R., & Copher, R. (2005). Families in the life course: Interdependency of roles,
role configurations, and pathways. Journal of Marriage and Family, 67(4), 858-
879.
McGowen, M. R., Ladd, L., & Strom, R. D. (2006). On-line assessment of grandmother
Minuchin, S., Lee, W., & Simon, G. M. (2006). Mastering family therapy: Journeys of
growth and transformation (2nd ed.). Hoboken, NJ: John Wiley & Sons, Inc.
Miller, R. B., Anderson, S., & Keala, D. K. (2004). Is Bowen theory valid? A review of
Mitchell, J., & Register, J. C. (1984). An exploration of family interaction with the
48-54.
Musil, C. M., & Standing, T. (2005). Grandmothers’ diaries: A glimpse at daily lives.
Musil, C. M., Warner, C. B., Zauszniewski, J. A., Jeanblanc, A. B., & Kercher, K. (2006).
61B(2), S89-S98.
Nichols, M. P., & Schwartz, R. C. (2001). The essentials of family therapy. Boston: Allyn
and Bacon.
119
Ohio’s Grandparents Raising Grandchildren’s Task Force. (1999). Grandparents raising
http://aging.ohio.gov/resources/publications/grandparents.pdf
Papero, D. V. (1990). Bowen family systems theory. Boston: Allyn and Bacon.
Peters, C. L., Hooker, K., & Zvonkovic, A. M. (2006). Older parents' perceptions of
Prest, L. A., & Protinsky, H. (1993). Family systems theory: A unifying framework for
Price, S. J., McKenry, P. C., & Murphy, M. J. (2000). In S. J. Price, P. C. McKenry, & M.
J. Murphy (Eds.), Families across time: A life course perspective (pp. 2-23). LA,
Pruchno, R., & McKenney, D. (2000). Living with grandchildren: The effects of
120
Raina, J.L. (1989). Structural and functional changes in the joint family system: A study
Robinson, M. M., & Wilks, S. E. (2006). "Older but not wiser": What custodial
grandparents want to tell social workers about raising grandchildren. Social work,
33(2), 164-177.
Rosenblatt, P. C. (1994). Metaphors of family systems theory. New York: The Guilford
Press.
Ross, C. E., & Mirowsky, J. (1988). Child care and emotional adjustment to wives'
Ross, M. E. T., & Aday, L. A. (2006). Stress and coping in African American
grandparents who are raising their grandchildren. Journal of Family Issues, 27(7),
912-932.
1069-1081.
Sandel, T. L., Cho, G., E., Miller, P. J., & Wang, S. (2006). What it means to be a
121
Sands, R. G., & Goldberg-Glen, R. S. (2000). Factors associated with stress among
Science Daily. (2008, November 4). Grandmothers as caregivers can cut risk of
Silverstein, M., & Ruiz, S. (2006). Breaking the chain: How grandparents moderate the
55(5), 601-612.
Sweet, J., Bumpass, L., & Call, V. (1988). The design and content of the National Survey
Sweet, J., & Bumpass, L. (2002). The National Survey of Families and Households -
122
Taylor, J. Y., Washington, O. G. M., Artinian, N. T., & Lichtenberg, P. (2007). Parental
stress among African American parents and grandparents. Issues in Mental Health
(Ed.), Clinical applications of Bowen family systems theory (pp. 51-68). New
U.S. Census Bureau. (2003). Grandparents living with grandchildren: 2000. Retrieved
http://www.census.gov/prod/2003pubs/c2kbr-31.pdf
U.S. Census Bureau. (2005). Who’s minding the kids? Child care arrangements: Winter
http://www.census.gov/prod/2005pubs/p70-101.pdf
U.S. Census Bureau. (2006). 65+ in the United States: 2005. Retrieved September 30,
209.pdf
U.S. Census Bureau. (2007). Annual Estimates of the Population by Sex and Five-Year
Age Groups for the United States: April 1, 2000 to July 1, 2007 (NC-EST2007-
01). Retrieved March 14, 2009 from the World Wide Web:
http://www.census.gov/popest/national/asrh/NC-EST2007-sa.html
U.S. Census Bureau. (2009). The 2009 Statistical Abstract. Retrieved May 16, 2009 from
123
Vandell, D. L., McCartney, K., Owen, M. T., Booth, C., & Clarke-Stewart, A. (2003).
Variations in child care by grandparents during the first three years. Journal of
Welsh, W.M., & Stewart, A.J. (1995). Relationships between women and their parents:
Wheelock, J., & Jones, K. (2002). 'Grandparents are the next best thing': Informal
childcare for working parents in urban Britain. Journal of Social Policy, 31(3),
441-464.
124
Appendix A: Descriptions of Constructs
125
Nights spent with GC
1) X1: During the last 12 months, about how many nights altogether, if any, did
1) X2: During last 12 months, about how often did you see your grandchild/any of
2) X3: During last 12months, about how often did you talk on the telephone or
3) X4: On a scale is from 0 to 10, where 0 is “not at all close” and 10 is "extremely
close," how would you describe your relationship with your grandchild/any of
Grandparent’s well-being
1) Y1: Happiness (RT201) - How would you say things are these days?
126
a. Scale – 1 : Very unhappy ~ 7: Very happy
b. Questions – On how many days during the past week did you…
ii. Not feel like eating or your appetite was poor? (RT206B)
iii. Feel that you could not shake off the blues even with help from
127
iv. Have trouble keeping your mind on what you were doing?
(RT206D)
b. Statements
for me (RT217D)
128
vii. I live life one day at a time and don’t really think about the future
(RT217G)
viii. I have not experienced many warm and trusting relationships with
others (RT217H)
ix. When I look at the story of my life, I am pleased about how things
(RT218B)
(RT218E)
life (RT218G)
xvi. Some people wonder aimlessly through life, but I am not one of
them (RT218H)
xvii. For me, life has been a continuous process of learning, changing,
129
xviii. I judge myself but what I think is important, not by the values of
Cultural attitude
3) C3: Adult kids with problems can live with parent (RT1809C)
Grandparent-parent support
1) GPPS1: Gave emotional support to child during the last month (RG5)
2) GPPS2:During the last month, helped children with childcare while they are
working (RG7)
3) GPPS3:During the last month, helped children with childcare while they are not
working (RG9)
130
5) GPPS5:Received help from adult children with housework, yard work, car repairs,
or other work around the house during the last month (RG14)
6) GPPS6:Received emotional support from children during the last month (RG16)
Other observed variables influencing the closeness between the grandparent and the
grandchild
2) Age of grandchild
131
b. GCO: Age of oldest grandchild (RH11) – Continuous variable
2) Grandparent’s ethnicity (recoded from the variable M484 from Wave 1 data)
a. Scale – 1: Yes, 2: No
132
Appendix B: Covariance Matrixes Used for Hypotheses Testing
133
No. GC GP Age GCY GCO X1 X2 X3 X4
21.308
17.669 108.365
-.189 11.413 10.905
24.607 56.839 8.893 69.634
3.010 -60.698 -4.715 -23.479 2142.693
.525 -1.839 -.655 -.375 16.163 1.956
.245 -1.554 -.213 -.072 13.605 1.022 2.321
.276 -.609 -.262 .400 13.239 1.328 1.319 4.131
Table B.1. Covariance Matrix Used for Hypothesis Test 1
X1 X2 X3 X4 Y1 Y2 Y3 Y4
2204.993
16.215 1.949
13.562 1.007 2.346
15.208 1.389 1.348 4.181
-3.680 -.006 .028 .313 1.521
-47.906 -.446 .120 2.721 7.012 84.627
-72.059 -1.291 -.695 -.544 9.209 67.427 246.400
-12.579 -.093 .422 .846 2.797 24.184 38.857 36.476
Table B.2. Covariance Matrix Used for Hypothesis Test 2
134
X1 X2 X3 X4 Y1 Y2 Y3 Y4 C1 C2 C3 GPPS FC
2215.040
16.103 1.960
13.450 1.012 2.355
15.472 1.407 1.368 4.246
-3.380 .004 .037 .310 1.517
-49.642 -.501 .124 2.616 7.083 85.408
-68.059 -1.283 -.644 -.641 9.335 68.526 250.437
-11.096 -.083 .412 .784 2.786 24.441 39.224 36.126
2.108 .104 .097 .146 -.008 -.346 -.800 .250 .871
.516 .051 .060 .094 .015 .098 -.510 .429 .327 .661
2.746 .094 .112 .100 -.053 -.443 -.504 -.013 .192 .126 .889
14.147 .875 .766 .894 -.135 -2.091 -4.178 .056 .171 .129 .148 2.546
-3.298 -.137 .023 .068 .790 7.968 16.018 4.324 .191 .193 .055 -.345 7.375
Table B.3. Covariance Matrix Used for Hypothesis Test 3
135
X1 X2 X3 X4 Y1 Y2 Y3 Y4 C1 C2 C3 GPPS FC
2569.547
17.168 1.781
13.920 .913 2.319
14.472 1.100 1.081 3.267
-3.475 -.001 .044 .287 1.649
-53.329 -.513 .056 2.502 7.911 94.842
-79.105 -1.303 -.785 .112 11.283 82.594 296.521
-16.677 -.035 .328 .798 3.203 27.198 49.023 39.696
3.483 .144 .124 .150 -.019 -.665 -1.351 .235 .861
1.041 .079 .100 .103 .011 .154 -.672 .482 .363 .627
3.804 .141 .151 .089 -.078 -.613 -.782 -.101 .179 .135 .906
14.216 .769 .644 .622 -.145 -2.135 -4.004 -.107 .233 .119 .200 2.432
1.139 -.004 .116 .259 .921 8.380 18.545 5.051 .182 .210 -.018 -.256 8.197
Table B.4.1. Covariance Matrix Used for Hypothesis Test 4 (Grandmothers)
X1 X2 X3 X4 Y1 Y2 Y3 Y4 C1 C2 C3 GPPS FC
1607.474
12.848 2.169
10.740 1.060 2.266
14.015 1.726 1.598 5.478
-2.555 .054 .077 .435 1.282
-38.713 -.182 .621 3.427 5.592 68.986
-26.779 .159 1.401 1.020 5.504 40.816 153.230
.811 .001 .762 1.100 2.030 19.393 20.588 29.996
-.133 .041 .054 .145 .009 .175 .082 .273 .888
-.623 -.013 -.028 .044 .029 .059 .006 .372 .270 .714
1.258 .032 .069 .149 -.017 -.214 -.284 .104 .212 .114 .861
10.647 .844 .703 .916 -.025 -1.350 -1.315 .695 .071 .110 .098 2.274
-8.202 -.208 .060 .060 .509 6.819 9.602 2.863 .203 .192 .150 -.159 5.790
Table B.4.2. Covariance Matrix Used for Hypothesis Test 4 (Grandfathers)
136
X1 X2 X3 X4 Y1 Y2 Y3 Y4 C1 C2 C3 GPPS FC
1412.925
14.551 2.012
12.971 1.038 2.411
15.064 1.605 1.575 4.693
-2.377 .026 .081 .368 1.425
-46.532 -.572 .163 3.220 7.111 83.703
-72.632 -1.225 -.727 1.006 9.507 67.556 223.974
-17.876 -.252 .235 .508 2.962 25.697 40.421 35.228
.978 .127 .061 .127 .024 -.017 -.673 .366 .870
.244 .071 .015 .061 .049 .468 .215 .485 .313 .666
3.347 .126 .081 .065 -.072 -.515 -.772 .260 .214 .127 .858
9.082 .807 .642 .864 -.209 -2.823 -4.595 -.255 .159 .136 .155 2.387
-2.730 -.065 .101 .287 .859 8.765 14.591 4.294 .318 .300 .126 -.275 7.734
Table B.5.1. Covariance Matrix Used for Hypothesis Test 5 (Caucasian American
Grandparents)
X1 X2 X3 X4 Y1 Y2 Y3 Y4 C1 C2 C3 GPPS FC
5013.658
23.759 1.861
19.918 1.039 2.482
16.901 .738 1.027 2.491
-6.916 -.059 .135 .132 1.627
-89.750 -.784 .931 1.387 6.976 101.136
-46.861 -1.517 1.698 -2.739 7.304 92.711 364.672
-1.119 -.756 .486 -.074 2.166 20.701 58.806 40.748
-1.652 .104 .101 .007 -.076 -.457 -.428 -.185 .798
.847 .041 .108 .023 -.084 -.613 -.646 1.025 .157 .487
-1.140 .047 .101 .088 -.071 -.060 .175 -.240 .211 -.010 .971
30.079 .732 .896 .422 -.209 -.965 -3.890 .223 .207 .064 .274 2.733
9.157 .000 .315 -.509 .510 6.287 30.937 8.138 .126 .099 .290 -.575 10.929
Table B.5.2. Covariance Matrix Used for Hypothesis Test 5 (Non-Caucasian American
Grandparents)
137
X1 X2 X3 X4 Y1 Y2 Y3 Y4 C1 C2 C3 GPPS FC
9607.934
35.954 1.721
19.861 .920 2.124
22.827 .666 .779 2.306
-3.457 -.061 .115 .174 2.018
-181.295 -2.203 -.811 1.315 10.059 116.307
-161.880 -1.673 -.249 -2.047 12.228 90.763 325.349
18.303 -.267 .927 .264 2.823 27.305 50.841 42.348
5.560 .074 -.032 .002 -.032 .337 -1.903 .455 .850
5.024 .047 .064 -.012 .130 .778 .149 .558 .342 .579
2.435 .124 .135 .098 .004 -.528 -1.427 -.760 .070 .091 .926
27.533 .782 .667 .395 -.300 -5.341 -6.730 -.095 .182 .088 .265 2.792
31.285 .385 .441 .759 1.348 10.278 28.556 6.228 .153 .364 -.206 -.424 11.081
Table B.6.1. Covariance Matrix Used for Hypothesis Test 6 (Custodial Grandparents)
X1 X2 X3 X4 Y1 Y2 Y3 Y4 C1 C2 C3 GPPS FC
808.360
10.488 1.972
9.708 .995 2.357
10.689 1.486 1.415 4.501
-1.238 .041 .054 .373 1.412
-10.127 -.001 .529 3.166 6.400 78.725
-17.209 -.783 -.210 .253 8.473 61.616 231.760
-8.494 .039 .432 1.009 2.700 23.295 35.968 34.843
1.028 .103 .111 .161 .002 -.411 -.524 .237 .874
-.564 .047 .054 .105 .000 .018 -.557 .421 .324 .674
2.579 .087 .105 .096 -.059 -.409 -.312 .118 .211 .132 .884
9.214 .857 .745 .926 -.077 -1.303 -3.254 .188 .161 .131 .125 2.469
-4.361 -.165 .019 .041 .647 7.163 13.111 3.834 .210 .173 .103 -.266 6.660
Table B.6.2. Covariance Matrix Used for Hypothesis Test 6 (Noncustodial Grandparents)
138