University of Southern California Dissertations and Theses

Childlessness and psychological well-being across life course as manifested in significant life events

(USC Thesis Other)

Childlessness and psychological well-being across life course as manifested in significant life events

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
CHILDLESSNESS AND PSYCHOLOGICAL WELL-BEING
ACROSS THE LIFE COURSE
AS MANIFESTED IN SIGNIFICANT LIFE EVENTS

by

Echo Win-Hu Chang
________________________________________________________

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(GERONTOLOGY)
May 2008

Copyright 2008 Echo Win-Hu Chang

ii

TABLE OF CONTENTS

List of Tables iii

Abstract

iv
Chapter One: Introduction

1
Chapter Two: Paper A
Childlessness and Psychological Well-Being of
Middle-Aged Caregivers

27

Chapter Three: Paper B
Childlessness and Psychological Well-Being of
New Retirees

68
Chapter Four: Paper C
Childlessness, Care Provisions, and the Psychological
Well-Being of Older Adults with Disability

106
Chapter Five: General Conclusion and Discussion 137

Bibliography

143
Appendix
Appendix A: Measures of Psychological Well-Being in HRS

161

iii

LIST OF TABLES

Table A1. Demographic Characteristics of the Sample 53
Table A2. Sociodemographic Characteristics by Parental Status and
Gender
54
Table A3. Characteristics of Caregivers by Gender, Marital Status,
and Parental Status
55
Table A4. Main Effects of Logistic Regression of Caregiver Burden
on Sociodemographic Characteristics of caregivers
56
Table A5. Main and Conditional Effects of Depressive Affect
Regressed on Parent–Child Relationship Among
Middle-Aged Caregivers and Noncaregivers
58
Table A6. Main and Conditional Effects of Positive Affect
Regressed on Parent–Child Relationship Among Middle-Aged
Caregivers and Noncaregivers
60
Table B1. Demographic Characteristics of Respondents 92
Table B2. Mean Scores for Depressive and Positive Affect 94
Table B3. Effects of Parental and Grandparental Status on
Depressive Affect in the Transition to Retirement
95
Table B4. Effects of Parental and Grandparental Status on Positive
Affect in Transition to Retirement
97
Table B5. Variables Affecting the Psychological Well-Being of Men
and Women
99
Table C1. Demographic Characteristics, ADL Difficulty, and ADL
Care Provision
129
Table C2. Logistic Regression on Assistive Device Use and Personal
Care
130
Table C3. Logistic Regression of Psychological Well-Being on
Demographic Characteristics of Older Adults With
Difficulty
132
iv

ABSTRACT
The childless population in the United States is growing fast, accounting for
approximately one fifth of population aged 65 and older. The combination of longevity,
marital status changes, and childlessness has led to the projection that about 25% of the
population aged 70 to 85 in 2030 will not have a living spouse or a living child.
Surprisingly, there is relatively little documentation about the lives of childless elders.
Childless people have been conspicuously ignored in social sciences, even in very
pertinent fields such as adult development, aging, the life course, and the family.
Among previous studies, two common shortcomings make results on
childlessness and old-age well-being inconclusive: (a) treating childlessness as a static
status and not considering possible changes in its meaning and impact felt in significant
life events, and (b) focusing on negative aspect of childlessness at the exclusion of
possible positive rewards. This dissertation added contextual variables and measured both
negative and positive affects of psychological well-being. It adopts stress process model
as a framework to study the effects of childlessness on psychological well-being in the
events of caregiving in middle age, retirement in young-old age, and disability in
advanced old age.
This dissertation uses data from Health and Retirement Study. The findings
indicate that childlessness by itself does not pose a negative threat to psychological well-
being. The lack of a negative effect of childlessness on psychological well-being does not
mean that children are not beneficial to parents. Rather, it suggests that taking the
―deviant‖ path of childlessness, whether voluntarily or not, does not necessarily render
v

childless persons at a greater disadvantage than parents in old age. However, one natural
outcome of childlessness is the lack of biological grandchildren the benefits associated
with grandparenthood. In this dissertation, grandchildren are found to have a positive
influence on grandparents‘, and in particular grandfathers‘, transitions to retirement.
The effects of childlessness were teased out from the impact of marital status.
Possible interaction of gender, parental status, and marital status were tested.
Childlessness and singlehood interacts in some cases. Childlessness poses different
implications for men and women.

1

Chapter One: Introduction
My dissertation, which adopts Pearlin and colleagues‘ stress process model
(Pearlin, Lieberman, Menaghan, & Mullan, 1981) as a framework, investigates (a)
whether children serve as a resource in mediating negative mental health outcomes of
stressful events at different stages of life, and thus (b) whether childless persons lack an
important resource with consequences to their psychological well-being. The dissertation
is composed of three papers under one overarching theme: the relationship of
childlessness and subjective psychological well-being in later life. Each paper explores
this relationship in one specific context, namely caregiving in middle age, retirement
transition in young-old age, and disability in very old age.
Empirical research suggests that the family is the primary source of social support
for elders and that adult children are the center of the resource network. Adult children
are regarded as sources of emotional, physical, and financial support for elderly parents
(Brubaker, 1990; Connidis & McMullin, 1994; McMullin & Marshall, 1996). Childless
people, particularly childless single women, suffer higher risks of institutionalization than
people with children (Boaz & Muller, 1994; Freedman, 1996). Therefore, childless elders
are often assumed to be more vulnerable than older parents and to suffer more
psychologically when facing old age because of the lack of offspring in their support
network (N. G. Choi, 1994). Contrary to this common assumption, most research has
reported no difference in the subjective psychological well-being of older parents and
childless elders. However, some research has shown that childless people may feel
insecure about their old age (Rubinstein, 1987). Does this insecurity materialize in
stressful events and manifest with negative effects on psychological well-being?
2

Surprisingly little is known about parental status and psychological well-being among
persons experiencing stressful events, such as the assumption of a caregiving role, the
transition to retirement, and disability. This dissertation sheds light on this issue.
Although most research has reported no difference in the subjective psychological
well-being of older parents and childless elders, researchers still cannot conclude with
certainty how or whether childlessness affects subjective psychological well-being
(Zhang & Hayward, 2001). This is due in part to methodological limitations in previous
studies, such as the use of nonrepresentative samples, cross-sectional comparisons, and
neglect of contextual factors such as the socioeconomic status of the participants.
Furthermore, previous research often treated the impact of childlessness as static, not
potentially fluctuating as life needs emerge. This dissertation overcomes some of these
limitations and extends upon previous studies. Its central objective is to study the effects
of childlessness on psychological well-being from middle age to very old age in the
context of significant life events that are prevalent at different life stages (i.e., caregiving
in middle age, retirement in young-old age, and disability in advanced old age).
People who experience similar life events and their associated stresses often have
different mental and physical health outcomes. The variation in outcomes may be better
understood by examining the resources available to people as buffers against stressors, as
proposed by Pearlin and colleagues (1981) in their stress process model. In addition, as
Pearlin and colleagues (1981) pointed out, it is a mistake to assume that a stressful event
automatically leads to negative outcomes or to adverse effects only. People are adaptive
and constantly work to maintain homeostasis; positive outcomes can come out of
stressful events. Similarly, although being childless has conventionally been viewed as a
3

disadvantage, it may pose some advantages, such as personal growth, strength in
autonomy, and sense of mastery of life (Cain, 2001), that may manifest in psychological
status. To capture the full range of psychological well-being, this dissertation measures
both positive and negative affect of subjective psychological well-being.
Background
Growth of the Childless Older Population
The childless population in the United States is growing fast. In 1990,
approximately one fifth of population aged 65 and older was childless (Himes, 1992).
Aging childless baby boomers will make childless couples one of the fastest growing
demographic segments in the coming two decades (Cain, 2001). The combination of
longevity, marital status changes, and the more common phenomenon of childlessness
has led to the projection that about 30% of the US Caucasian population aged 70 to 85 in
2030 will not have a living spouse or a living biological child (Wachter, 1997).
However, surprisingly, there is relatively little documentation about the lives of
childless elders (e.g., their heath, social service utilization, and general well-being; Zhang
& Hayward, 2001). The research and literature on childlessness is scant (Cain, 2001;
Jefferies & Konnert, 2002). Childless people have been conspicuously ignored in social
sciences, even in very pertinent fields such as adult development, aging, the life course,
and the family (Dykstra & Hagestad, 2007b). The lives of childless people are an
understudied area (K. R. Allen, Blieszner, & Roberto, 2000).
Meaning of Childlessness Throughout History
Parenthood anchors the ―normal expectable life‖ (Dykstra & Hagestad, 2007b, p.
1276). The centrality of parenthood is reflected in how people conceptualize the life
4

course. For example, a young married couple may describe having babies as ―starting a
family,‖ as if households without babies are not legitimate families, and children growing
up and leaving home is often a central theme in middle age (Neugarten, 1968).
To be sure, the meaning of childlessness has changed along with social changes in
history. It used to be seen as deviant; childless people were viewed either with pity for
not being able to reproduce due to health problems, or with revulsion and given the label
―anti-children‖ if they openly admitted to being voluntarily childless. Now that social
norms have changed, childlessness has gained more acceptance and tolerance and is now
even regarded as a lifestyle choice among younger cohorts (May, 1995; Thornton, 1989).
The change in social attitudes toward childlessness is reflected in national survey data. In
1962, 84% of mothers who participated in the Study of American Families agreed with
the statement that ―almost all married couples who can, ought to have children‖; by 1980,
only 43% of mothers in the same study expressed agreement with the statement
(Thornton & Young-DeMarco, 2001). As Cain (2001) noted, there is a generation gap
with regard to how women feel about childlessness. According to Cain‘s qualitative
study, childless women in their 50s (those born in the 1950s) acknowledge a strong
societal imperative toward having children that makes them feel inadequate for not
conforming to the norm. In contrast, childless women in their 40s (those born in the
1960s) have mixed reactions about social pressure and their status as childless. Some feel
a societal expectation was placed on them, whereas others do not. Childless women who
are younger than 35 do not feel an obligation to motherhood and thus feel more
comfortable with their childlessness. The next generation could well be free to be
childless without caring about societal sanctions.
5

However, even though childlessness is more accepted, it is still not regarded as a
desirable option by most people. For example, in 1988, only 6% of married women
between 18 and 44 expressed the intention to remain childless, according to data from the
U.S. General Social Survey (Dykstra & Hagestad, 2007b). Being childless may still
represent a chronic stressor for childless people in a pronatalist society such as the United
States that lauds the benefits of having children (E. Campbell, 1985) and punishes
childlessness through public opinion and tax benefits for parents (Cain, 2001). It may be
even more stressful for older adults, as their socialization may have imposed a stronger
stigma for deviating from social norms stronger than that experienced by younger
generations. This may manifest as negative impact on psychological well-being.
In addition, being childlessness may imply a person is missing important benefits
in life (e.g., emotional insurance, a sense of immortality, and a sense of children as
surrogate achievements), which are also common motivations for parenthood (E.
Campbell, 1985). Children and grandchildren can provide family members with a sense
of security, affection, and continuity. Without these benefits, childless elders may face
double the insecurity: fear of no future help, and inadequate real-time support. These can
impose a heavy burden on childless people‘s psychological well-being (Zhang &
Hayward, 2001).
Evolution of Research on Childlessness and Aging
The Early Days (Pre-1970s)
Years ago, it was widely believed that the risk of loneliness for childless older
persons was indiscriminant, widespread, and persistent (Houser, Berkman, & Beckman,
1984). This negative view on childlessness was compounded by traditional beliefs about
6

the importance of adult children as sources of emotional, financial, and instrumental
support, which gave rise to concerns over the psychological well-being of childless elders
(Zhang & Hayward, 2001). This negative notion about lacking offspring in a person‘s life
was reflected in early research that focused mostly on the negative aspects of being
childless, and studies on psychological well-being measured depressive aspects only.
Bias in treating childlessness as ―abnormal‖ was also evident in lifespan
developmental theories. The experience of parenthood was viewed as an important
developmental task (Havighurst, 1972), which was also clear in family life cycle theories
that proposed that life stage progression is defined by the development of the first child.
Parenthood was viewed as a biological imperative (Gutmann, 1975), the proper
development of self, behavior (i.e., masculine or feminine behavior), and personality
were considered to be shaped by, and in response to, the needs of children in their
different developmental stages. Although he did not equate generativity—or a concern
with supporting and guiding the next generation—with parenthood, Erikson (1968) in his
stages of development theory did suggest that being a parent is perhaps the most common
means of expressing generativity.
Meanwhile, it was assumed that childless elders were disadvantaged. This
assumption stemmed from the observation that childless people have smaller social
networks, and so thus must have less social contact and less social support. Under such
assumptions, early researchers often adopted a problem-oriented perspective and looked
at childlessness as inherently stressful rather than considering both the pros and cons of
not having children or adopting a normative–adaptive perspective that directed attention
to substitutive resources childless persons cultivate for themselves in the process of
7

adapting to their childless situation.
Subsequent No-Difference Findings From the Research
In spite of these problem-oriented perspective and theories, research findings
continually showed no differences between parents and childless persons in subjective
psychological well-being. Almost all studies reported that, contrary to conventional
belief, childlessness was not directly linked to subjective psychological well-being.
Childless elders, most of them older widows, did not conform to the stereotype of
unhappiness and dissatisfaction (Beckman & Houser, 1982; Houser et al., 1984;
Koropeckyj-Cox, 1998, 2002; Zhang & Hayward, 2001). In short, although it only
measured the negative side of psychological well-being, research consistently found that
childlessness did not confer a statistically significant level of greater loneliness or
depression.
This no-difference conclusion was partially explained by compensatory support
that childless people cultivate throughout life and mobilize in old age. Because
childlessness is a long-term status, people with no offspring have time to build up their
support reserves with relatives and friends. Family members other than offspring often
serve as sources of support for childless people. Research has found that childless persons
see more of their nieces and nephews than parents do; they also depend more on siblings,
nieces, and nephews than do elderly parents (N. G. Choi, 1994; Wenger, Scott, &
Patterson, 2000). Childless persons are fully integrated in social networks and receive
needed support (Connidis & McMullin, 1992; McMullin & Marshall, 1996). As reported
by McMullin and Marshall, support from family may buffer the impact of stress for
parents, whereas support from friends may work more for elderly childless persons than
8

for aging parents. Therefore, support networks composed of relatives and nonkin may
work for childless persons as adult children do for elderly parents.
However, the perception that childless elders are more vulnerable persisted
through time, supported by the fact that these people do have fewer support sources
(Dykstra, 2006; Furstenberg, 2005; Mugford & Kendig, 1986; Wenger et al., 2000), and
nonprimary relations, such as friends, neighbors, and distant kin, are unlikely to commit
to providing long-term care (Jerrome & Wenger, 1999; Johnson & Catalano, 1981;
Wenger, Dykstra, Melkas, & Knipscheer,, 2007). For example, Dooghe (1994)
considered childless persons one of the specific risk groups among elders, drawing this
conclusion from a United Nations research project on population aging. The argument
was that close family ties, such as those with spouses or children, cannot be easily
replaced by other relationships. Researchers did find that spouses and children provided
better and more intensive care to elders (Chapman, 1989), which, consequentially, may
have caused differences in psychological well-being between those with spouses and/or
children and those without. The current trend among the older population of utilizing
more formal care to substitute for or supplement informal care has ushered in the
question of whether the gap in care quality and quantity, if indeed there is such a gap,
will narrow in the future for parents and childless persons.
Differentiating Between Involuntarily ―Childless‖ and Voluntarily ―Childfree‖
In an attempt to make sense of the conflicting evidence comparing parents and
childless elders in old age well-being, some scholars came to the conclusion that the
inconsistency in findings was partially due to indiscrimination of intragroup differences.
In other words, previous research had ignored the diversity within groups. It was
9

suggested that the quality of the parent–child relationship among parent groups, and the
reasons for being childless (e.g., voluntary, circumstantial, or medical), may be linked to
psychological status (Connidis & McMullin, 1993; Jeffries & Konnert, 2002;
Koropeckyj-Cox, 2002).
For example, with regard to diversity among parents, both Connidis and
McMullin (1993) and Koropeckyj-Cox (2002) found that the typology of parental status
was important—a good quality parent–child relationship may represent an important
source of social and emotional support and pride and pleasure in being a parent (and
possibly a grandparent), whereas a strained parent–child relationship may pose a threat to
the psychological well-being of the aging parents. Their findings corresponded to
Silverstein and Bengtson‘s (1991) report that emotionally close parent–child relationships
diminished loneliness and depression in parents‘ old age. In contrast, if the relationship
between parents and offspring is poor, it may evoke a sense of failure and constant stress
and negatively impact the psychological well-being of parents (Koropeckyj-Cox, 2002).
Similar differentiation could be applied to the childless group. Whereas voluntarily
childfree couples might express more satisfaction with life and greater self-determination
than parental groups (E. Campbell, 1985), some stress of coping with involuntary
childlessness may persist or reemerge in later life (Beckman & Houser, 1982; Connidis &
McMullin, 1993). Researchers found that persons who classify themselves as voluntarily
childless might view childlessness as a desired lifestyle, successfully develop their own
support networks throughout their life course (Andrews, Abbey, & Halman, 1991;
Callan, 1987; Connidis & McMullin, 1993), and hold a similar level of psychological
well-being as parents who have a close relationship with their offspring. In addition,
10

studies found that involuntarily childless persons were less happy and more depressed
than emotionally close parents (Connidis & McMullin, 1993) but not worse off than
parents with poor offspring relationships (Koropeckyj-Cox, 2002).
No Need to Differentiate Childless and Childfree?
The differences between childfree and childless groups render it important to
differentiate the origin of childlessness. However, the division between voluntary and
involuntary childlessness is not always clear. The distinction is often the result of
untreated infertility or a succession of events that have ultimately resulted in having no
children (Kemkes-Grottenhaler, 2003). As Cain (2001) observed from her qualitative
research on childless women, although some women actively choose childlessness and
some tragically wind up without children against their wish, there are a great many
women in between. Some women unexpectedly evolve into childlessness; they are
childless not by intention but by circumstance, such as late marriage, financial
constraints, a focus on the career, and so on. The reaction of these women to their status
as childless may change with time or even with relationships; for example, a woman may
regard herself as voluntarily childless if she is married to a divorced man with children
from a previous marriage who does not want to father any more children even though, if
left to her own will, she would prefer to mother biological children. Therefore, the
division between voluntary childless by choice and involuntary childless by circumstance
is often blurred.
The difficulty in classifying childless groups was illustrated by Connidis and
McMulllin (1996). They used a self-identification method to place participants in
childless by choice or childless by circumstance groups and asked participants to state
11

their reason for being childless. The researchers reported that there was a 60% overlap in
the specific reasons given for childlessness among these two groups that theoretically had
distinctly different reasons for not having had children.
The different ways of classifying the voluntary or involuntary origin of
childlessness are partially responsible for discrepancies in findings among studies
(Jeffries & Konnert, 2002), as some researchers classified the participants according to
their own criteria, while others allow participants to be self-classified. Zhang and
Hayward (2001) stated that adaptation may reduce the differences of being childless by
choice and by circumstance, thus making it less meaningful to differentiate the origin of
childlessness. It has also been proposed that, regardless of whether a person is voluntarily
or involuntarily childless, if he or she adapts well, outcomes in psychological well-being
may be similar, as life progresses and childlessness takes a backseat to a comfortable
routine, personal fulfillment, and marital harmony (E. Campbell, 1985).
In summary, the literature is inconclusive about whether childlessness leads to
loneliness and depression and worse psychological well-being among elders. On the one
hand, research has consistently found no differences among aging parents and childless
groups; on the other hand, the no-difference findings seem to be contingent upon marital
status, gender, and relationship quality (Zhang & Hayward, 2001). I propose that, besides
demographic characteristics and social relationships, researchers should also consider the
context variable—the ―situation‖ or ―event‖ in which psychological well-being is
measured. The perception of one‘s parental status may change at different points of time,
conditioned by the emergence of needs. Taking into consideration the situational factor
adds one extra dimension in assessing the dynamics between parental status and
12

psychological well-being. This dissertation, which takes into account the event in
assessing the relationship between parental status and psychological well-being, has the
potential to more fully tease out the relationship between childlessness and psychological
well-being in later life.
What Researchers Know so Far
Parental Status and Well-Being
Easterlin (2003) attempted to explain what contributes to people‘s feelings of
well-being as expressed in social surveys. He concluded that the most important factors
contributing to subjective well-being are (a) material living level and (b) marital status
and health level (almost equally important). This was in accordance with Beckman and
Houser‘s (1982) report that marital status, financial status, and health are key factors in
determining people‘s psychological well-being in old age. Research has found that
whether one has children or not does not have a significant influence on one‘s subjective
well-being. However, most research has tested parental status and well-being in general.
The question remains whether parental status makes a difference when psychological
well-being is threatened by adverse events or stressors—a time when one most needs
social support.
Age Effect on Psychological Well-Being in the Context of Parental Status
Age by itself does not change the level of psychological well-being of a childless
person. Research findings have suggested few age differences in the experience of being
childless in terms of regretting not having children (Jeffries & Konnert, 2002).
Nevertheless, Alexander, Rubinstein, Goodman, and Luborsky (1992) argued that lack of
children should have its greatest negative effect on persons when they are aging, which is
13

when their physical, financial, and social resources diminish. However, their study did
not differentiate the marital, financial, or health status of the respondents, and the
authors‘ hypothesis had little support, as social gerontologists have generally reported
that childless older women neither are less satisfied nor have lower well-being than other
older women (Beckman & Houser, 1982; Zhang & Hayward, 2001). Beckman and
Houser found that unless a person experiences widowhood, low income, and bad health,
having no children in old age does not affect well-being. This implies that the negative
impact of childlessness is adversity ascribed, not age specific. However, one does face
higher odds of widowhood and health deterioration when in old age. Therefore, not
having children may correlate with, although not cause, poorer psychological well-being
in older persons.
Parental Status in the Context of Marital Status
Zhang and Hayward (2001) suggested that, because it is inconclusive whether
childlessness has any bearing on psychological well-being, it may be more revealing if
the question is put in the context of marital status and gender. This makes sense, as
childless married persons tend to rely on each other as primary sources for fulfilling most
of their needs (Johnson & Catalano, 1981) and do not feel as though they are missing
support when they are married. The implications of their having a smaller social support
network than parents may not hit home until widowhood or divorce occurs. Is a single,
childless elder more vulnerable than a parent when the childless person does not have
spousal support? This question is answered to a good extent in existing research as
summarized in the following paragraphs.
A study by Beckman and Houser (1982) indicated that childlessness has a greater
14

negative effect on well-being for widowed than married persons. Widowed women have
lower psychological well-being than married women. However, the study also showed
that even in widowhood, childlessness has only a very minimal effect on well-being;
other factors, such as physical capacity, religiosity, quality of social interaction, and
strength of social support, moderate the effect. However, a qualitative study by Wenger
(1997) of people aged 80 and older reported that newly widowed childless men and
women expressed a wish at that moment that they had had children. In contrast to
Wenger‘s report, Zhang and Hayward (2001) found that little evidence supports the idea
that childlessness has negative effects on psychological well-being among widowed,
divorced, and never-married persons. They further stated that there is a consensus in the
literature that childless persons are not more likely to be lonely or depressed compared
with parents and stepparents once marital status is controlled. The discrepancy among
these studies could be partially due to the fact that Beckman and Houser and Wenger
used nonrepresentative samples and Wenger used qualitative methods, whereas Zhang
and Hayward conducted a quantitative study with a nationally representative sample.
Upon divorcing, fathers may become more alienated from their children than
mothers are, because most children in the United States remain with their mothers after
marital disruption. The increased divorce rate is likely to have a great impact on the
provision of informal care to elderly fathers. But in general, both divorced fathers and
mothers are less likely than married parents to receive care from their children due to
emotional strains in their children‘s own lives (Cicirelli, 1983; Mugford & Kendig,
1986). This may have a bearing on the psychological well-being of divorced elders.
In contrast, the situation of never-married persons is unique. Traditionally,
15

children come after marriage. Among the current generation of elders, it was not socially
acceptable to have children out of wedlock. Therefore, these individuals were not
inclined to be fathers or mothers if they were not married. Researchers long ago reached a
consensus that never-married persons utilize strategies to cultivate their support networks
throughout their lives because they have no spouse or offspring to rely on when the need
for care arises (Johnson & Catalano, 1981).
Parental Status in the Context of Gender
Zhang and Hayward (2001) rightly pointed out that research on childless elders
and late-life depression should take into account the context of gender. Due to different
socialization and role expectations, men and women may ascribe different meanings to
parenthood and childlessness. For example, in her study of psychological well-being
among voluntarily and involuntarily childless men and women, Koropeckyj-Cox (2002)
found that childless men did not report diminished well-being, but childless women, and
particularly involuntarily childless women, did so when compared with mothers who had
a close relationship with their offspring. Motherhood has traditionally been given more
emphasis as a meaning maker in women‘s lives than fatherhood has been in men‘s
(Veevers, 1973). Childlessness may impact women more negatively than men due to
gender socialization and the ostensible imperative of motherhood in women‘s lives
(Blake, 1979; Fisher, 1991). It is not uncommon to hear women reporting feeling
stigmatized due to their status as childless (Cain, 2001; Miall, 1986; Veevers, 1980).
Questions of identity and value may persist into old age or resurface in old age for
women (Alexander et al., 1992). It is noticeable that there is more emphasis on studying
women than men regarding feelings on childlessness. One reason, as stated previously, is
16

that conventional thinking attributes a more significant impact to women than to men
with regard to being without offspring (Broverman, Vogel, Broverman, Clarkson, &
Rosenkrantz, 1972; Russo, 1976; Veevers, 1980). Another is that the conjunction of
childlessness with widowhood, which is assumed to be a dire situation for receiving care,
is more common for women than for men. Because men are less likely to be affected by
widowhood, they are not as often studied as women are. The result is a dearth of findings
on childlessness among men (Bulcroft & Teachman, 2003), starting with rarely available
data on men‘s fertility (Greene & Biddlecom, 2000). The lack of interest in childless men
among researchers reflects a typical research model of childbearing and motherhood
(Forste, 2002; Greene & Biddlecom, 2000).
Conjunction of Marital Status, Parental Status, and Gender
Do childless widows experience more depression than childless widowers? A
study by Koropeckyj-Cox (1998) did not support this notion. Koropeckyj-Cox found that
for both men and women, widowhood was linked with significantly greater loneliness
and generally greater depression regardless of parental status. But Zhang and Hayward
(2001) did find a gender difference in loneliness and depression among widowed
childless persons. They reported that widowed childless men were 1.74 times more likely
than widowed childless women to feel lonely as well as depressed. Dykstra and Hagestad
(2007a) also found that childlessness makes more difference in men‘s lives than women‘s
in terms of social embeddedness and the social contact it entails, particularly for formerly
married men.
The Need for This Dissertation
There are two main reasons why it is important to understand how childlessness
17

affects the aging experience: Childlessness (a) is the status of approximately 20% of the
older population and (b) has implications for aging policy and public services.
Four societal forces further point to the importance of understanding aging among
childless people, and thus the need for this dissertation:
1. The childless population is fast increasing and now comprises 20% of
baby boomers (Himes, 1992);
2. Dramatic changes have taken place in family structures in recent decades
that have influenced the provision of filial support;
3. Changes in social norms are lending different meanings of childlessness to
different cohorts; and
4. Much of the research on childless people‘s psychological well-being dates
back to the early 1980s; existing findings may not be applicable to cohorts
currently entering old age, and even less so to baby boomers
(Goldscheider, 1990; Koropeckyj-Cox, 1998; Preston, 1992). Therefore,
there is a need to continuously monitor the meaning and impact of
childlessness on the aging population, to gauge the group‘s well-being,
and to project demand for public assistance in old age.
Expected Contribution of This Dissertation to Advancing Knowledge
Does childlessness matter to psychological well-being? From the literature
review, the answer is not conclusive. Some limitations in previous research possibly
contribute to this inconclusiveness. These limitations include the following:
1. Use of small and nonrepresentative samples;
2. Neglect of antecedent variables such as race, gender, and social class;
18

3. Overlooking of potential mediating factors such as marital status and
gender;
4. Examination of only the negative dimension of psychological well-being;
and
5. No consideration of contextual factors such as stressful events that
demand social support that is often provided by family members,
particularly adult children.
To overcome these limitations, this dissertation draws on data from the Health
and Retirement Study (HRS), a nationally representative longitudinal study.
Sociodemographic variables such as marital status, gender, race, income, and health were
controlled to isolate the effects of childlessness. To ensure a balanced examination, I
measured both positive affect and depressive affect of psychological well-being. Finally,
each of my three studies focused on one stressful event—caregiving, retirement, and
disability—in order to discern the impact of childlessness on psychological well-being
when a person experiences one of these stressful events but does not have children in the
center of his or her support network.
Although childlessness may have few effects on the onset of psychological
problems, it may exact a toll in terms of greater risk of institutionalization and death
(Kendig, Dykstra, van Gaalen, & Melkas, 2007). Therefore, childlessness potentially
comes into play at different points in the aging process (Zhang & Hayward, 2001). This
supports the need to study the relationship between childlessness and psychological well-
being in different stages across late life in the context of stressors prevalent in those
stages, as this dissertation does.
19

Another unique feature of this dissertation is that it recognizes that not having
children also means not having biological grandchildren. In view of the increased
importance of grandparenthood in modern times, this dissertation has a relatively more
complete assessment of the impact of childlessness with the inclusion of
grandchildlessness.
This dissertation helps bring up to date the understanding of the relationship
between childlessness and the aging process. Furthermore, it can serve as a reference for
policy planning aimed at maintaining the boomer generation‘s well-being in old age,
while providing insight into the adequacy of social support for the childless population
and these people‘s potential reliance on formal services.
Methods
Data Source
This dissertation uses data from different waves of the HRS, conducted by the
Institute for Social Research at the University of Michigan. In each paper of this
dissertation, I analyze a subset of HRS data meeting the sampling criteria of that
particular paper. The sample selection process for each paper is described in detail in that
respective paper.
The HRS is a longitudinal study with a household panel surveyed every 2 years.
The original sample for the first study (1992) consisted of respondents aged 51 to 61 at
the time of survey and their spouses (regardless of the spouse‘s age). The total number of
respondents was 12,652 in 1992. In 1998, its companion study, the Study of Asset and
Health Dynamics Among the Oldest Old (AHEAD), was integrated into the HRS;
together they are the revised HRS. AHEAD had an initial sample of 7,447 respondents
20

aged 70 and older (and their spouses, regardless of the spouse‘s age). The HRS and
AHEAD started to use a uniform questionnaire after they merged in 1998.
The HRS collects data on demographic characteristics, health status, kin
networks, familial resources, care arrangement, and other domains. It is suitable for
examining how family structure (i.e., not having children and/or grandchildren) impacts
the psychological well-being of persons in later life, which is the central concern of this
dissertation.
Definition and Operationalization of Key Constructs Used Across Papers in This
Dissertation
Definition of Childlessness
In this dissertation, childlessness is defined as the status of never having had a
biological child, regardless of whether one has an adopted child or stepchild. This is the
most widely adopted definition of childlessness in aging research (Koropeckyj-Cox,
1998). I chose to limit the definition of childlessness to biological children because of
data constraints in the HRS that do not provide information on adopted children, and also
because of the uncertainty associated with the stepparent–stepchild relationship. Because
the HRS does not survey whether a person has adopted any children, studies using HRS
data have no means of exploring this particular form of parent–child relationship
(Koropeckyj-Cox, 1998; Zhang & Hayward, 2001). Information about stepchildren is
available in the HRS but was excluded from this dissertation because of the inherent
ambiguity and diversity so common in the experience of stepparenthood (Cherlin &
Furstenberg, 1994) that stepchildren are not viewed as a reliable source of support for old
age as biological children. Therefore, the existence or absence of stepchildren is not
21

considered in this dissertation. It is worth noting that Zhang and Hayward suggested that
stepchildren can be a potential source of social support, and Koropeckyj-Cox (1998)
raised the issue that most surveys do not make distinctions between biological children,
adoptive children, and stepchildren, which is indeed a question that deserves greater
attention. However, national data on adopted children and stepchildren are rare. The most
complete and recent data are from U.S. Census 2000 and show that 2.5% of all children
related to householders are adopted and 5.2% are stepchildren (Kreider, 2003). These
numbers can serve as indicators of how many householders who have children are
adoptive parents or stepparents (which can be in addition to their status as biological
parents), but not of the percentage of exclusively adoptive parents or stepparents among
the adult population.
In this dissertation, permanently childless adults are compared to their peers who
are biological parents with at least one living child. This dissertation does not include
those parents who survived all of their children, due to the assumption that the meaning
of parenthood is different for a person who once had children but lost them all and a
person who never had biological children. It is likely that parents who have outlived all of
their children have gone through life with the identity of mother or father; childlessness
for these people is unexpected and has not always been a part of their lives as it has for
other childless persons. However, the number of parents who survived all of their
children is small and accounts for only 0.9% (230 out of 27,031 participants in the HRS
panel from 1995–2004). The percentage of parents who outlived all of their children is
reported in each paper for each of the three studies, respectively.
Operationalization of Childlessness
22

The HRS asks respondents about their parental status when they are first
surveyed. The question is not put to them in the successive waves because the majority of
participants are aged 50 and older (unless the respondent is a much younger spouse of the
primary respondent) and are assumed to have passed the age of childbearing. The
question is worded as follows: ―How many children have you fathered/have you given
birth to? Please don‘t count miscarriages or still-births, or adopted or step-children for
this question.‖
Because the HRS is a panel study, parental status information for most
respondents is recorded in the early waves (mostly in the 1995 AHEAD and 1996 HRS
waves; this question was not asked in the 1992 or 1994 HRS waves), with later waves
containing parental status information for new participants only. I combined data from all
waves to form a complete list of the number of biological children of each participant.
The question regarding whether a respondent has ever had a biological child is
followed by one that surveys the number of living children: ―How many of them are still
living?‖ As with the question about having had a biological child, the question about the
number of living children is asked the first time a participant joins the panel but not again
in successive waves. This poses a potential problem of being unable to identify
respondents who have lost all of their children after their first wave of participation in the
HRS. However, the number of parents who outlive their children is small (less than 1%
of the entire HRS sample); therefore, this limitation should not impact my study
significantly.
To compile a complete set of parental status information, I compared the number
of biological children and the number of living children. If the number of biological
23

children was at least 1, and the number of living children was not 0, then the person did
not meet the criterion of outliving all children, and I recorded the outliving-all-children
variable as 0. If the person did not have a biological child, there was no question about
surviving all children, so the outliving-all-children value was also 0. Only when a person
had biological child(ren) and no living child was he or she counted as having outlived all
children and the value for that variable entered as 1. A person who survived all children
was excluded from the sample for this dissertation, as explained in ―Definition of
Childlessness.‖
There are two limitations to operationalizing the childlessness construct as
described in the preceding paragraph. First, I was unable to discriminate the origin of
childlessness—whether it was voluntary by choice or involuntary by circumstance.
However, as explained in ―Evolution of Research on Childlessness and Aging,‖ the line
between voluntary and involuntary childlessness is often blurred, and adaptation
practiced by people through time may make the distinction less meaningful with regard to
outcome. Second, as explained previously, because the question on number of living
children is only asked once when a participant first joins the HRS panel, I had no way of
knowing whether the person suffered the loss of any child(ren) in subsequent waves that
would have made him or her meet the criterion of having outlived all children. However,
because less than 1% of the entire HRS sample outlived all of their children, this should
not be a serious concern.
Definition of Psychological Well-Being
Analogous to the idea that health is not merely the absence of disease,
psychological well-being is not merely the absence of depression. Psychological well-
24

being is a multidimensional concept that acknowledges the potential coexistence of both
positive and negative feelings. In the past two decades, studies on how to measure
psychological well-being have consistently found that positive and negative affect are
two dominant yet relatively independent dimensions (Watson, Clark, & Tellegen, 1988).
Therefore, when assessing psychological well-being, researchers should investigate the
manifestation of both positive and negative aspects of emotions.
Almost all previous research on the psychological well-being of childless elders
or widowed older persons has focused on negative aspects (i.e., depression and
loneliness). This dissertation expands the investigative scope to include positive aspects
(i.e., happiness and enjoyment of life). Furthermore, it is a mistake to assume that
seemingly stressful events always lead to stress (Pearlin et al., 1981), or that stressors
necessarily lead to negative outcomes only. Instead, both positive and negative outcomes
should be measured. For example, caregiving, although demanding, can bring joy in the
helping of a loved one and can be linked to a subjective feeling of well-being (Stuckey,
Neundorfer, & Smyth, 1996). At its outset, retirement, which ushers a person into
uncharted territory, may cause stress, but it also involves anticipation and many benefits.
Less positive projections can be made about disability, but social support, including that
from adult children, may change or mediate the process and outcomes.
Three indicators of psychological well-being—depression, loneliness, and
sadness—composed the measure of depressive affect in my study. They have
traditionally been cited in normative admonitions against childlessness (Koropeckyj-Cox,
2002). Loneliness has been defined as an unpleasant feeling of dissatisfaction with either
the number or the quality of existing social relationships (DeJong-Gierveld, 1987;
25

Koropeckyj-Cox, 1998, 2002; Perlman & Peplau, 1982). Depression is ambiguous in
definition and has various manifestations of feeling low, blue, or simply down (Blazer,
1995). The manifestation of positive aspects of well-being is recognized in the form of
global happiness, satisfaction with family life, and satisfaction with friendship, which are
measured as ―being happy‖ and ―enjoying life‖ in the Center for Epidemiologic Studies–
Depression scale (CES-D; Radloff, 1977).
Operationalization of Psychological Well-Being
Psychological well-being is measured in the HRS with the CES-D, a self-report
depression scale for research in the general public, which was designed to measure
depressive symptoms by emphasizing their subjective and affective elements (McDowell
& Newell, 1996; Radloff, 1977). The CES-D has been widely used in research on the
general and clinical populations (Radloff, 1977), and it has good internal consistency
(Hann, Winter, & Jacobsen, 1999). The HRS adopted a 9-item battery of CES-D items,
with each item limited to a dichotomous yes/no response. Six of the nine items indicate
the presence of depression as manifested in negative feelings and somatic symptoms, and
three indicate positive feelings. Positive feelings include being happy, enjoying life, and
having a lot of energy much of the time during the past week; negative feelings include
being depressed, lonely, or sad much of time during the past week; somatic indicators
include feeling that everything is effort, sleeping restlessly, and finding it hard to get
going. The lead-in question for the CES-D in the HRS reads as follows: ―Now think
about the past week and the feelings you have experienced. Please tell me if each of the
following was true for you much of the time during the past week. Much of the time
during the past week you felt depressed. (Would you say yes or no?)‖ Additional eight
26

questions to measure psychological well-being are worded in the same format. Appendix
A contains the nine items that are presented to respondents. Several published papers
have used the CES-D to study mental health based on HRS data (Gallo, Bradley, Siegel,
& Kasl, 2000; Siegel, Bradley, Gallo, & Kasl, 2003, 2004), and Gallo and colleagues
concluded that it has acceptable measurement properties after assessing its reliability and
validity using 1992 and 1994 HRS data.
For this dissertation, individual CES-D items were grouped into two indexes:
depressive affect and positive affect of psychological well-being. I first used factor
analysis to extract a positive affect factor and a depressive affect factor from the nine
items. The factor analyses extracted two factors: (a) depressive affect, which included the
three original CES-D items of being depressed, feeling lonely, and feeling sad; and (b)
positive affect, which included the two original CES-D items of being happy and
enjoying life. Next I added together items belonging to the same factor to form an index
of depressive affect (0–3 items) and positive affect (0–2 items).
Although I initially transformed the original discrete dichotomous measure into an
ordinal measure through factor analysis and index-composition process, I eventually
treated depressive affect and positive affect as a dichotomy for the logistic regression
analysis. In the regression, depressive affect was either 0 (no depressive symptoms at all)
or 1 (one to three depressive symptoms). Positive affect was coded as 0 (not being happy
and not enjoying life) or 1 (being happy or enjoying life, or both).

27

Chapter Two: Paper A
Childlessness and Psychological
Well-Being Among Middle-Aged Caregivers

28

Introduction

The life course perspective suggests that people generally assume social roles in
line with age norms in the context of social and historical time (Elder & Johnson, 2003).
Social time refers to a normative concept that specifies an appropriate age for
transitioning to different life stages and assuming new roles. Historical time is the period
that affects personal development and links a person with a cohort. Currently, middle-
aged people have a greater chance of assuming a caregiver role than do people in younger
or older generations, and they will act as caregivers for a longer period of time than have
past caregivers. This is due to the developmental stages of people close to them, such as
their parents, senior relatives, and friends, who are experiencing longevity but need
assistance to maintain community living. In light of the longevity typically enjoyed by
today‘s older generation, people who are entering middle age may have aging parents in
need of care, or even an aging spouse. On top of this, baby boomers, currently between
44 and 61 years old, collectively delayed childbearing, which makes them more likely to
be sandwiched between caring for aging parents and raising children. As the trend shows,
the number of women older than 30 who have given birth has risen steadily since 1976
(National Center for Health Statistics, 1990). Consequently, there is an increased
likelihood that baby boomers will experience the overlap of child care and elder care
responsibilities (Himes, 1992).
But the recipients of elder care are not limited to one‘s aging parents. They could
be relatives, friends, and neighbors. Due to changes in family structure from a pyramid to
a present-day beanpole shape, with each generation in the nuclear family having fewer
members, extended family members have become increasingly important in providing
29

social support for nuclear family members. Fictive families, whose members are related
not by blood but by choice, have become increasingly common, too, particularly among
African Americans and the gay and lesbian communities (Barker, 2002). Consequently, a
person may enjoy the resources provided by relatives, friends, and neighbors from
extended and fictive family and, correspondingly, take on the task of caring for those not
in the immediate family. Therefore, the caregiving dyad is not limited to those in a
parent–child relationship. As Barker (2002) noted, neighbors, friends, and other nonkin
caregivers were important for sustaining community living for about 10% of the frail
elders in her sample. The variegated support networks popular today contribute to
making caregiving a common experience for middle-aged men and women.
People in middle age may juggle more roles than people in any other age group
while engaging in multiple domains such as the family, work, education, caregiving,
community services, and so on. But the caregiver role is not like other roles. First and
foremost, it is a role that few plan to assume. Caregiving is often described as an
―unexpected career‖ (Pearlin & Aneshensel, 1994), thus regardless of the age or life stage
of the caregiver, even though it appears to be an expected social role for middle-aged
persons. As a result, assuming the caregiver role requires adjustment and a redistribution
of resources to each role in a person‘s role repertoire.
Becoming a caregiver requires being both available and willing to provide care.
The literature has shown that middle-aged adults with fewer family demands are more
likely to provide care to frail or sick individuals because they are more available; for
example, daughters and sons who are single are more likely to help parents, and for
longer periods of time, than are those in couples (Laditka & Laditka, 2001). This is
30

attributed to the fact that they do not commonly face competing demands from spouses
and in-laws. This signifies the fact that marital status impacts caregiver availability. Less
is known about parental status and caregiving. Are childless people more available, and
thus more likely to become caregivers, because they do not have a parental role? There is
almost no research on the relative odds of childless people and parents providing non-
child care.
Caregiving research has mostly focused on negative outcomes and paid
considerably less attention to the possible positive impact of caregiving, although in
reality caregiving satisfaction and caregiving burden often coexist in persons caring for a
loved one (Stuckey, Neundorfer, & Smyth, 1996). Even faced with the stressful event of
caring for a spouse or parent afflicted with Alzheimer‘s disease, caregivers express some
sort of satisfaction with their experiences (Lawton, Moss, Kleban, Glicksman, & Rovine,
1991). In light of the scarcity of research on parental roles and caregiving, plus the
attention focusing on negative outcomes of caregiving, it is not surprising that few studies
were found that provide data on the potential positive effects of parental role on
caregivers.
Furthermore, researchers know little about the implications of sparse family
structures (i.e., the lack of a spouse or offspring) on the psychological well-being of
caregivers. Do caregivers who are single and childless have less stress because they have
fewer role conflicts? Or do they have more stress because they have no spouse or
children as support resources? What is the impact of offspring on middle-aged
caregivers‘ psychological well-being when caregiver role and parental role are put in the
context of gender and marital status? The intersection of marital role, parental role, and
31

caregiver role on the psychological well-being of caregivers has not being fully explored
in the literature.
Theoretical Framework
This study adopts Pearlin and colleagues‘ stress process model (Pearlin,
Lieberman, Menaghan, & Mullan, 1981) as a framework to investigate whether children
serve as a resource in mediating the negative mental health outcomes or enhancing the
positive mental health outcomes of the stressful event of caregiving. The impact of
parental status is studied within the context of gender, marital status, and other social
relationships.
Research has well documented that long-term caregiving often imposes burden and
strains on caregivers (Pearlin, Pioli, & McLaughlin, 2001). But not all caregiving entails
stress, and sometimes negative outcomes are accompanied by positive rewards for some
caregivers (Pearlin, Lieberman, Managhan, & Mullan, 1981). The variation in individual
outcomes may be explained to a good extent by the social support at the command of the
caregiver, in addition to other factors such as the caregiver‘s individual personality. Both
Pearlin and colleagues (1981) in their stress process model and Seeman and Crimmins
(2001) in their serial studies of social environments and health identified social support as
a resource to mediate the negative physical and mental outcomes of caregiving. Social
support may also have protective effects on physical health by reducing the likelihood of
the biological stress response (Seeman & Crimmins, 2001; Umberson, Williams, &
Sharp, 2000), which, in turn, should have a bearing on psychological well-being.
In addition to stress process model, this study adopts the lens of social role theory
to examine how one‘s parental role, in conjunction with other social roles, affects
32

caregiving outcomes. Two competing social role theories attempt to explain the
relationship between social role and psychological well-being. One of them is the role
strain approach, which proposes that the greater number of roles one assumes, the greater
the likelihood of role incompatibility and demands, thus the greater the role strain and
conflict (Burr, Leigh, Day, & Constantine, 1979; Goode, 1960). According to role strain
approach, the overwhelming demand of multiple roles eventually produces psychological
distress. In contrast, the role enhancement approach posits that performing multiple roles
can be rewarding enough to outweigh the negative effects of demands from multiple roles
(Marks, 1977; Sieber, 1974); thus, having multiple roles may be negatively associated
with psychological distress.
How do the parental and caregiver roles interact with each other, in conjunction
with other social roles, to affect the psychological well-being of middle-aged caregivers?
Both the stress process model and social role theory can shed light on this issue.
Background
Family and Caregiving Support
Family is the most important social support source for the majority of people.
Family structure and its functioning influence caregiving experiences for caregivers
(Mitrani et al., 2006). Researchers have also defined caregiving as an experience
affecting the whole family, not just the caregiver. Caregiving affects other family
members through diverting a caregiver‘s time away from those family members,
changing the family lifestyle, producing feelings of guilt in caregivers, and so on (Brody,
Hoffman, Kleban, & Shoonover, 1989; Gatz, Bengtson, & Blum, 1990; Strawbridge &
Wallhagen, 1991; Toseland, Smith, & McCallion, 1995). If negative interaction with
33

other family members occurs, it may become for the caregiver a secondary stressor in
addition to the primary stressor of caregiving (Pearlin, Lieberman, Menaghan, & Mullan,
1981).
In light of the aforementioned research findings, it appears that what matters for
caregivers is not only their composition of roles, but also the quality of relationships
embedded in those roles. In other words, other relationships in a caregiver‘s life can serve
as supportive forces or competing demands. For example, Voydanoff and Donnelly
(1999) tested the interaction between caregiving and marital quality in relation to
depressive symptoms and found that daughter caregivers experienced fewer depressive
symptoms if they enjoyed a high-quality marriage. Choi and Marks (2006) extended the
study design to be longitudinal and reported that caregivers caring for biological parents
and spouses experienced more negative consequences in the presence of high marital
disagreement. Apparently, high-quality marriage has a buffer effect against negative
caregiving outcomes. Extending this thinking along the line of parental roles, it is
reasonable to expect that not only the existence of offspring, but their relationship to the
caregiver, influences parent caregivers.
With regard to how parental and caregiver roles combine to affect caregivers‘
psychological well-being, research findings are not consistent. Some studies have
reported no relationship between caregiving and psychological well-being for parents
with children at home (Loomis & Booth, 1995; Stoller & Pugliesi, 1989; Voydanoff &
Donnelly, 1999). However, Stull, Bowman, and Smerglia (1994) found that children
either have a positive effect on caregivers by reducing caregivers‘ sense of strain in
certain domains of their lives, or they have no effect and do not contribute to caregivers‘
34

sense of strain. In their study, Stull, Bowman, and Smerglia did not find that the number
of children (ranging from 0 to 5 or more) in the household constituted a competing
demand for the caregiver. According to the authors, combining the caregiver role with
other roles does not negatively impact psychological well-being. These findings are in
line with role enhancement theory as opposed to role strain theory. However, these
studies measured only the negative dimensions of psychological well-being and neglected
potential positive outcomes. They also did not take the quality of the parent–child
relationship into consideration.
In contrast, research has provided consistent evidence that having children in the
home reduces global happiness and poses negative effects on parents‘ psychological
well-being (Campbell, Converse, & Rodgers, 1976; Glenn & McLanahan, 1981; Glenn &
Weaver, 1979). A number of studies have demonstrated that having children at home
increases demands, economic hardship, strains in marriage, and workload in general,
which may lead to feelings of depletion and negatively affect parents‘ psychological
well-being (Gove, 1984; Ross & Huber, 1985; Ross, Mirowsky, & Goldsteen, 1990). In
addition, members of the ―sandwich generation‖ commonly experience negative
outcomes as a result of combining parental and caregiver role. In short, research findings
on the relationship between the parental and caregiver roles are inconclusive when it
comes to having underage children at home and nonexistent for nonresident adult
children: The question remains whether offspring serve as a resource for parents who
assume a caregiver role.
The argument in favor of children as resources suggests that offspring represent the
continuity of generations, thus giving the parent–caregiver a sense of security in
35

projecting potential future help for himself or herself when he or she needs it. Perceived
availability of help, defined as the sense of having someone who cares and who will help
if one is in need, is beneficial to health and well-being in later life (Norris & Kaniasty,
1996). If this is indeed so, then, does childlessness represent a deficit in resources for
caregivers? Are childless caregivers at greater risk for negative outcomes in caregiving?
Or, as role strain theory suggests, does childlessness result in less distraction and
psychological distress because it represents one less role competing for caregivers‘ time?
Gender Differences in Support for Caregivers
Conventional wisdom regards caregiving as a woman‘s job. Women still shoulder
the majority of caregiving (Gerstel & Gallagher, 1993; Miller, 1990: Spitze, Logan,
Joseph, & Lee, 1994; Stoller, 1990), but men are stepping up as an equally important part
of caregiver pool. In fact, for baby boomers, the caregiving responsibility is less gender
divided. For example, in a survey of multicultural baby boomers, AARP (2001) found
equal proportions of men (22%) and women (22%) aged 45 to 55 providing care to
parents, other aging relatives, or both.
One might speculate that, because women in older cohorts were socialized to be
nurturing and family oriented whereas men were expected to be devoted to jobs and other
external responsibilities, male caregivers face more problems than female caregivers
(Pruchno & Resch, 1989). However, Wallsten (2000), in her study of older couples,
found that female caregivers had more negative feelings about their husbands than male
caregivers had about their wives even though female caregivers had more social supports
than male caregivers. This is complemented by another study that found that male
caregivers had more favorable indicators of psychological stress. Some caution is due in
36

interpreting the findings, because a potential bias in response may be at play here. One
should take into consideration the different ways of expressing feelings between the
genders, and specifically the fact that women are more open in expressing distress.
Gender differences in caregiving outcomes may be partially due to life-long gender
differences in psychological well-being and not entirely attributable to the caregiving
situation (Walker, Pratt, & Eddy, 1995).
However, a study by Zarit, Reever, and Bach-Perterson (1980) found no
differences in feelings of burden among husband and wife caregivers. Moreover, male
and female caregivers showed no differences in the use of formal or informal support in a
meta-analysis by Pinquart and Sörensen (2006), which integrated 229 studies on gender
differences in caregiver psychological well-being, physical well-being, and social
resources.
Although men and women may have similar social resources, they may draw
different levels of comfort from interactions with their children due to the possible
different meanings of motherhood and fatherhood. Motherhood has been said to mean
more to women than fatherhood means to men (Broverman, Vogel, Broverman,
Clarkson, & Rosenkrantz, 1972; Russo, 1976; Veevers, 1980). Therefore, support from
their children, or a lack of it, may carry different weight for mothers and fathers. For
example, Marks (1995), in his study of mid-life marital status and relationships with adult
children, reported that social support relationships with adult children had more of an
impact on the well-being of mothers than fathers. This finding is in line with the
proposition that women‘s psychological well-being is more contingent on relationships
than is men‘s (Gilligan, 1982; Jordan, Kaplan, Miller, Stiver, & Surrey, 1991; Kessler &
37

McLeod, 1984). This may imply possible gender differences in the effects of support for
caregivers among spouses and children. Moreover, it is a conventional belief that family
roles, such as spouse, parent, and caregiver, are more salient for women than men; the
conjunction of these roles may have a stronger effect on women‘s psychological well-
being than on men‘s.
Research Questions
Are children a resource that mediates caregiving outcomes, or are they a distraction
vying for caregivers‘ resources? These two competing speculations highlight the need for
understanding the impact of the presence of child(ren) on middle-aged caregivers. In
addition, whether and how the parent–child relationship impacts caregivers‘
psychological well-being is not clear. In view of the findings on marital quality and
gender differences and their relation to caregiver well-being, the impact of parental role
should be studied alongside these factors. Furthermore, the potential positive impact that
children have on parent–caregivers has been neglected in previous research, which points
to the need for investigating psychological well-being in both its negative and positive
spheres.
This paper extends previous research on parental and caregiver roles by using
parent–child closeness to predict the odds of caregiver burden and of positive and
negative emotions associated with caregiving among middle-aged persons. It focuses on
the following research questions:
1. Is being childless, being a parent with no close child, or being a
parent with a close child associated with self-reported caregiver burden among
middle-aged persons who provide frequent care to others?
38

2. Does being childless, being a parent with no close child, or being a
parent with a close child affect positive affect of psychological well-being of
middle-aged persons who provide frequent care to others?
3. Does being childless, being a parent with no close child, or being a
parent with a close child affect depressive affect of psychological well-being of
middle-aged persons who provide frequent care to others?
To isolate the effects of the parental role from those of the marital role, this study
also investigated how being never married, being previously married (i.e., separated,
divorced, or widowed), or being married is associated with self-reported caregiver
burden, negative emotions, and positive emotions among middle-aged persons who
provide frequent care to others.
Methods
Data Source and Sample Selection
Data for this paper came from the lifestyle version of the Leave-Behind
Questionnaire of the Health and Retirement Study (HRS), which surveyed a random
subsample of the 2004 HRS. The HRS is a longitudinal study with a household panel
surveyed every 2 years. The original sample for the first study (1992) consisted of
respondents aged 51 to 61 at the time of survey and their spouses (regardless of the
spouse‘s age). In 1998, its companion study, the Study of Asset and Health Dynamics
Among the Oldest Old (AHEAD), was integrated into the HRS; together they are the
revised HRS. AHEAD had an initial sample of 7,447 respondents aged 70 and older (and
their spouses, regardless of the spouse‘s age). The HRS and AHEAD began using a
uniform questionnaire after they merged in 1998.
39

In 2004, the HRS added a new feature of data collection in the form of self-
administered questionnaires (―Leave-Behind Questionnaires‖) that were left with
respondents upon the completion of an in-person core interview. The purpose of the
Leave-Behind Questionnaire was to collect additional information from respondents
without adding to the interview length. Two separate questionnaires, one on the
participant‘s lifestyle and the other on work and health, were left to separate subgroups.
The lifestyle questionnaire, which included questions on participation in general
activities, relationships with others, and views on life in general as well as specific
aspects of the respondent‘s life, contained a specific inquiry on the caregiving experience.
Data for this wave were collected from March 2004 through February 2005.
The HRS subsamples for the Leave-Behind Questionnaires were representative of
both the HRS population and the similarly aged U.S. population. The 2004 HRS sample
was first divided into two random half-samples, and then the subsample for each Leave-
Behind Questionnaire was randomly sampled from within one of the half-samples. The
lifestyle questionnaire was administered to respondents of all ages. The response rate was
around 80%.
There were 3,232 respondents to the lifestyle Leave-Behind Questionnaire, ranging
in age from 31 to 104 years old. Because my study focuses on middle-aged men and
women, I included only respondents who were between 49 and 64 years of age at the
time of the survey. This age range fits the common conception of middle age. It is also
safe to assume that after age 49, most childless persons will be permanently childless in
the sense of biological parenthood. The average age of menopause inception for
American women is approximately 51 years old (National Institute on Aging, 2007).
40

Statistics on the fertility and/or childlessness of men are much rarer (Greene &
Biddlecom, 2000), but men have significantly less ability to father a child after age 50.
For example, the British Office for National Statistics reported in 2006 that only 1 in 100
babies were born to a father older than 50 (this statistic includes all children born, not just
the first child; Beckett, 2006). By deduction, the odds of a man fathering his first child
after age 50 must be less than 1 in 100. Therefore, people who are older than 50 and who
are childless are likely to be permanently biologically childless. To increase the number
of childless people in the sample, I selected respondents as young as 49 years old. People
aged 49 and 50 accounted for 11.5% of the childless group in this sample. The upper age
limit was 64, because the conventional wisdom adopted by most aging researchers is that
middle age ends at 64 and old age starts at 65.
My final sample for analysis comprised 1,342 men and women aged 49 to 64.
Among them, 42.3% (n = 568) were caregivers and 57.7% (n = 774) were noncaregivers;
88.2% (n = 1,184) were parents and 11.8% (n = 158) were childless. The average age was
56.84, and the mean education level was 13.15 years.
Definition of Key Variables
Definition of childlessness
In this study, childlessness was defined as the status of never having had a
biological child, regardless of whether one has an adopted child or a stepchild. This is the
most widely adopted definition of childlessness in aging research (Koropeckyj-Cox,
1998). I chose to limit the definition of childlessness to biological children because the
HRS does not provide information on adopted children, and also because there may be
uncertainty associated with the stepparent–stepchild relationship. Because the HRS does
41

not ask whether a person has adopted any children, studies using HRS data have no
means of exploring this particular form of parent–child relationship (Koropeckyj-Cox,
1998; Zhang & Hayward, 2001). Information about stepchildren is available in the HRS
but was excluded from this study because of the inherent ambiguity and diversity so
common in the experience of stepparenthood (Cherlin & Furstenberg, 1994). Therefore,
the existence or absence of stepchildren was not considered in this study. It is worth
noting that Zhang and Hayward (2001) suggested that stepchildren can be a potential
source of social support, and Koropeckyj-Cox (1998) raised the issue that most surveys
do not make distinctions between biological, adoptive, and stepchildren, which is indeed
a question that deserves greater attention.
In this study, permanently childless adults were compared to their peers who were
biological parents with at least one living child. This study did not include those parents
who survived all of their children, due to the assumption that the meaning of parenthood
is different for a person who once had children but lost them and a person who never had
biological children. The number of parents who survived all of their children is small and
accounts for only 0.9% (230 of 27,031) of participants in the HRS panel from 1995 to
2004. In this study, only 8 participants (0.06% of the sample) who met the age
requirement for the sample had outlived their child(ren). They were excluded from the
final sample.
Operationalization of childlessness
The HRS asks respondents about their parental status when they are first surveyed.
The question is not put to them in the successive waves because the majority of
participants are aged 50 and older (unless the respondent is a much younger spouse of the
42

primary respondent) and are assumed to be past the age of childbearing. The question is
worded as follows: ―How many children have you fathered/have you given birth to?
Please don‘t count miscarriages or still-births, or adopted or step-children for this
question.‖
Because the HRS is a panel study, parental status information for most respondents
was recorded in the early waves (mostly in the 1995 AHEAD and 1996 HRS waves; this
question was not asked in the 1992 or 1994 HRS waves), with later waves containing
parental status information for new participants only. I combined data from all waves to
form a complete list of the number of biological children of each participant.
The question regarding whether a respondent has ever had a biological child is
followed by one that surveys the number of living children: ―How many of them are still
living?‖ As with the question about having had a biological child, the question about the
number of living children is asked the first time a participant joins the panel but not again
in successive waves. This poses a potential problem of being unable to identify
respondents who have lost all of their children after their first wave of participation in the
HRS. However, the number of parents who outlive their children is small (less than 1%
of the entire HRS sample); therefore, this limitation should not have a significant impact
on the study.
To compile a complete set of parental status information, I compared the number of
biological children and the number of living children. If the number of biological children
was at least 1, and the number of living children was not 0, then the person did not meet
the criterion of outliving all children, and I recorded the outliving-all-children variable as
0. If the person did not have a biological child, there was no question about surviving all
43

children, so the outliving-all-children value was also 0. Only when a person had
biological child(ren) and no living child was he or she counted as having outlived all
children and the value for that variable entered as 1. A person who survived all children
was excluded from the sample for this study, as explained in ―Definition of
Childlessness.‖
Definition of psychological well-being
Analogous to the idea that health is not merely the absence of disease,
psychological well-being is not merely the absence of depression. Psychological well-
being is a multidimensional concept that acknowledges the potential coexistence of both
positive and negative feelings. Therefore, when assessing psychological well-being,
researchers should investigate the manifestation of both positive and negative aspects of
emotions.
Almost all previous research on the psychological well-being of childless elders or
widowed older persons has focused on negative aspects (i.e., depression and loneliness).
This study expands the investigative scope to include positive aspects (i.e., happiness and
enjoyment of life). Furthermore, it is a mistake to assume that seemingly stressful events
always lead to stress (Pearlin, Lieberman, Menaghan, & Mullan, 1981), or that stressors
necessarily lead to negative outcomes. Both positive and negative outcomes should be
measured. For example, caregiving, although demanding, can bring joy in the helping of
a loved one. In the case of caring for parents, it tends to be demanding in terms of
incompatibility with other roles but also results in psychological and interpersonal
rewards (Voydanoff & Donnelly, 1999). Three indicators of psychological well-being—
depression, loneliness, and sadness—composed the measure of depressive affect in my
44

study. They are traditionally cited in normative admonitions against childlessness
(Koropeckyj-Cox, 2002). Loneliness has been defined as an unpleasant feeling of
dissatisfaction with either the number or the quality of existing social relationships
(DeJong-Gierveld, 1987; Koropeckyj-Cox, 1998, 2002; Perlman & Peplau, 1982).
Depression is ambiguous in definition and has various manifestations of feeling low,
blue, or simply down (Blazer, 1995). The manifestation of positive aspects of well-being
is recognized in the form of global happiness, satisfaction with family life, and
satisfaction with friendship in most adopted measures. In the Center for Epidemiologic
Studies–Depression scale (CES-D; Radloff, 1977), they are measured as ―being happy‖
and ―enjoying life.‖
Operationalization of psychological well-being
Psychological well-being is measured in the HRS with the CES-D, a self-report
scale that was designed to measure depressive symptoms by emphasizing their subjective
and affective elements (McDowell & Newell, 1996; Radloff, 1977). The CES-D has been
widely used in research on general and clinical populations (Radloff, 1977), and it has
good internal consistency (Hann, Winter, & Jacobsen, 1999). The HRS adopted a 9-item
battery of CES-D items, with each item limited to a dichotomous yes/no response. Six of
the nine items indicate the presence of depression as manifested in negative feelings and
somatic symptoms, and three indicate positive feelings. Positive feelings include being
happy, enjoying life, and having a lot of energy much of the time during the past week;
negative feelings include being depressed, lonely, or sad much of time during the past
week; somatic indicators include feeling that everything is effort, sleeping restlessly, and
finding it hard to get going. The lead-in question for the CES-D in the HRS reads as
45

follows: ―Now think about the past week and the feelings you have experienced. Please
tell me if each of the following was true for you much of the time during the past week.
Much of the time during the past week …‖ Several studies have used the CES-D to study
mental health based on HRS data (Gallo, Bradley, Siegel, & Kasl, 2000; Siegel, Bradley,
Gallo, & Kasl, 2003, 2004), and Gallo and colleagues concluded that it has acceptable
measurement properties after assessing its reliability and validity using 1992 and 1994
HRS data.
In this study, individual CES-D items were grouped into two indexes: depressive
affect and positive affect of psychological well-being. The following steps were involved
in the transformation: (1) using factor analysis to extract a positive affect factor and a
depressive affect factor from the nine items.; the factor analyses extracted two factors: (a)
depressive affect, which included the three original CES-D items of being depressed,
feeling lonely, and feeling sad; and (b) positive affect, which included the two original
CES-D items of being happy and enjoying life, and (2) adding together items belonging
to the same factor to form an index of depressive affect (0-3 items) and positive affect (0-
2 items).
Although I initially transformed the original discrete dichotomous measure into an
ordinal measure through factor analysis and index-composition process, I eventually
treated depressive affect and positive affect as a dichotomy for the logistic regression
analysis. In the regression, depressive affect was either 0 (no depressive symptoms at all)
or 1 (one to three depressive symptoms). Positive affect was coded as 0 (not being happy
and not enjoying life) or 1 (being happy or enjoying life, or both).
Caregiver
46

In this study, caregivers were identified by a nominal variable with caregiver coded
as 1 and noncaregiver as 0. Caregivers were identified through a scale inquiring about
caregiving experience. The question asks whether the respondent had ―help[ed] at least
one sick, limited, or frail family member or friend on a regular basis.‖ The four response
choices are (a) No, didn‘t happen; (b) Yes, but not upsetting; (c) Yes, somewhat
upsetting; and (d) Yes, very upsetting. I classified participants answering ―No, didn‘t
happen‖ as noncaregivers with a value of 0; respondents providing any of the other three
answers were regarded as having provided care to others and were counted as caregivers
with a value of 1 for the caregiver variable.
Dependent Variables
Caregiver burden
Caregiver burden is a multidimensional response to stress felt from the caregiving
experience, including physical, mental, financial, and time aspect. It is often defined as
the strain or load endured by a person who provide informal care to others for a long
period of time (Kasuya, Polgar-Bailey, & Takeuchi, 2000; Stuckey, Neundorfer, & Smyth,
1996). In this study, caregiver burden was determined by the self-reported state of feeling
upset with caregiving activities in the question used to determine caregiver status (i.e.,
whether the respondent had ―help[ed] at least one sick, limited, or frail family member or
friend on a regular basis‖). Caregivers who answered ―Yes, but not upsetting‖ about their
caregiving experience were regarded as not experiencing caregiver burden, whereas
persons who answered ―Yes, somewhat upsetting‖ or ―Yes, very upsetting‖ were
classified as experiencing caregiver burden.
Depressive affect
47

Depressive affect was a factor of three original CES-D items of being depressed,
feeling lonely, and feeling sad. The value was either 0 (no depressive symptoms at all) or
1 (one to three depressive symptoms).
Positive affect
Positive affect included the two original CES-D items of being happy and enjoying
life. It was coded as 0 (not being happy and not enjoying life) or 1 (being happy or
enjoying life, or both).
Predictive Variables
Parental status
In this study, I first differentiated participants‘ status as parents (n = 1,184) or
childless (n = 158), then I further classified the parent group into two groups according to
self-reported relationship with child(ren). If participants reported having a close
relationship with at least one child, then they were considered parents with a close child
(n = 1,140); if they reported having no close relationship with any of their children, then
they were considered parents with no close child (n = 44). In all, 96.3% of parents
reported having a close relationship with at least one child.
The need to differentiate the parent–child relationship is based on previous research
findings that suggest that the quality of the parent–child relationship may be linked to
psychological status (Connidis & McMullin, 1993; Jeffries & Konnert, 2002;
Koropeckyj-Cox, 2002). For example, with regard to diversity among parents, studies of
both Connidis and McMullin (1993) and Koropeckyj-Cox (2002) found that the typology
of parental status was important—a good quality parent–child relationship may represent
an important source of social and emotional support and pride and pleasure in being a
48

parent, whereas a strained parent–child relationship may pose a threat to the
psychological well-being of the aging parent. These findings corresponded to Silverstein
and Bengtson‘s (1991) report that emotionally close parent–child relationships appeared
to diminish loneliness and depression in parents‘ old age. In contrast, if the relationship
between parents and offspring is poor, it may evoke a sense of failure and constant stress
and negatively impact the psychological well-being of parents (Koropeckyj-Cox, 2002).
Similar differentiation could be applied to the childless group, whereas voluntarily
childless couples might express more satisfaction with life and greater self-determination
than parents (Campbell, 1985), while the stress of coping with involuntary childlessness
may persist or reemerge in later life (Beckman & Houser, 1982; Connidis & McMullin,
1994). Thus, it worthwhile to differentiate the origin of childlessness. However, due to
data constraints inherent in the HRS, it is impossible to differentiate childlessness as
voluntary or involuntary.
Marital status
In this study, participants‘ marital status was classified into three categories: (a)
married (i.e., married or coupled), (b) previously married (i.e., separated, divorced, and
widowed), and (c) never married. The small sample sizes for the childless group and
parents with no close child made it impractical to further classify the married or the
previously married groups. However, neither was it theoretically sound to group
previously married and never-married persons into one ―unmarried‖ group: Never-
married persons often adopt life-long strategies for developing their social support
network, whereas previously married persons may lose an important source of support
49

(the spouse) unexpectedly due to divorce or widowhood, which may affect their
psychological well-being.
I included self-reported couples in the married group because cohabitation may
provide some of the same benefits as marriage (Koropeckyj-Cox, 2007), including
intimacy, instrumental support, social integration, and economic advantages (Dykstra &
Hagestad, 2007a). In fact, the HRS 2004 Leave-Behind Questionnaire treated spouses
and partners in a similar fashion; its question about relationship was worded as follows:
―How close is your relationship with your partner or spouse?‖
Other Control Variables
Resident child(ren)
Three variables were used to group participants with regard to their children‘s
residential status. These variables were as follows: (a) being childless (yes = 1); (b)
having at least one child at home (yes = 1); and (c) having child(ren), but none of them
residing at home. But this variable was eventually dropped from the final model, due to
the desire for parsimony by foregoing variables which do not add explanation power to
the model.
Employment status
Combining the worker and caregiver roles has more of an impact on women‘s
lives than on men‘s due to the socialization of women and men in their expected roles.
Little attention has been paid to men‘s experiences with conflicting worker and caregiver
roles. Some have reported that working in conjunction with providing parental care and
child care leads to greater stress for female caregivers than for male caregivers (Older
Women‘s League, 1998; Stone, Cafferata, & Sangl, 1987). Others have found that
50

employment may be beneficial for female caregivers, as it may provide economic or
social benefits (e.g., interaction with, and support from, coworkers) or enhance self-
esteem (Nieva, 1985), which may ameliorate stress from caregiving. Regardless, the
relationship is more likely to be correlational than causal (Stull, Bowman, & Smerglia,
1994), meaning that in most studies the link between multiple roles and stress is assumed
but not demonstrated. In their study, Stull, Bowman, and Smerglia (1994) found no
support for the claim that employment reduces caregiver strain; rather, they concluded
that there is no relationship between employment status and social or financial strain, but
that caregiving does increase physical strain. In sum, findings on the effects of combining
the worker and caregiver roles on female caregivers‘ psychological well-being are
inconclusive; many of the outcomes may depend on individual caregiver characteristics
and resources, as proposed by the stress process model. Therefore, when studying
caregiving outcomes, it is essential to consider caregivers‘ sociodemographic
characteristics.
Race
It has been long established that there is a huge variation among races with regard
to the concepts of filial obligation and familism and support resources for caregivers
(AARP, 2001; Knight et al., 2002). Due to the small sample size of the childless group,
this study coded race simply as Caucasian or minority without further differentiating
ethnicity among the minority group. Caucasian was the reference group in this study.
Health , education level, and income
Income and education levels are highly correlated. Health and income are
significantly related to well-being in life (Easterlin, 2003). These factors may be
51

influential to caregivers‘ psychological well-being because they represent the resources
caregivers have, or lack, in performing caregiving tasks and other demanding social roles.
Other important social relationships
Personal relationships, including those with siblings, friends, and other family members,
are frequently regarded as source of aid among baby boomer caregivers across cultures
(AARP, 2001). For childless unmarried persons, other close relationships may be even
more salient as resources of support in stressful events. In the HRS Leave-Behind
Questionnaire, the question about close family members refers to immediate family
members other than the spouse or children, such as parents, brothers and sisters, cousins,
grandchildren, and so on. The values of the dichotomous variable ―having at least one
close family member or not‖ in my study are determined through two questions. The first
question asks ―Do you have any other immediate family, for example, any brothers or
sisters, parents, cousins or grandchildren?‖ then, with a follow-up question asking ― With
how many people in your immediate family would you say you have a close
relationship?‖ If the respondent reported to have at least one close family member, then,
he or she is giving a value of 1 (i.e., having at least one close family member); otherwise,
the value is 0 (i.e., having no close family member). A similar method was used to
determine whether a respondent has at least one close friend with questions regarding
friendship.
Statistical Analyses
I used factor analysis to extract two major factors from the nine CES-D items in the
HRS data. Only five items had significant weights on two factors. Therefore, only five
CES-D items were used for analysis. Three items (depression, loneliness, and sadness)
52

fell into one factor that represented depressive affect, and two items (happiness and
enjoying life) constituted another factor representing positive affect. The 3-item factor of
depressive affect had a reliability score of .776, and the 2-item factor of positive affect
had a reliability score of .717.
I ran descriptive statistics to gain a profile of the sample in terms of their
sociodemographic characteristics. Then I applied t tests to compare group differences in
demographic characteristics and behaviors associated with the caregiver role according to
gender, marital status, and parental status. Finally, I applied logistic regression to test the
odds of having caregiving burden, depressive affect, and positive affect on various
sociodemographic and interaction terms between the caregiver role, parental role, and
marital role. Two models of logistic regression for caregiver burden and three models for
depressive affect and positive affect, respectively, were tested in this study.

Results
Table A1 presents a profile of the sample. Approximately 42.3% were caregivers,
and 57.7% were noncaregivers. Among the entire sample, 40.2% participants were men
and 59.8% were women. The gender composition of the caregivers roughly corresponded
to that of the entire sample, with 38.2% men and 61.8% women. The majority of
participants were married or coupled (75.1%), Caucasian (83.7%), and had at least one
biological child (88.2%). Among the younger participants (i.e., baby boomers aged 58
and younger), the percentage of childless participants was slightly higher than in the
entire sample (13.8% vs. 11.8%). This reflected the fertility rate changes in cohorts.

53

Table A1. Demographic Characteristics of the Sample (n = 1,342)
Characteristic n %
Gender
Male 540 40.2
Female 802 59.8
Marital status
Married/coupled 1,008 75.1
Previously married
(separated, divorced, widowed)
279 20.8
Never married, single 55 4.1
Race
Caucasian 1,123 83.7
Minority 219 16.3
Parental status
Parent close child relation 1,140 84.9
Parent no close child relation 44 3.3
Childless 158 11.8
Caregiver
Caregiver 568 42.3
Noncaregiver 774 57.7
Currently employed
Yes 814 60.7
No 528 39.3

Table A2 presents comparison of demographic characteristics by parental status
and gender. Compared with mothers, childless women had higher education levels (p <
.001). Childless women were also more likely than mothers to participate in labor force
and be employed (p < .001). It is not surprising that there was a higher proportion of
married or coupled persons in the parent group, because parental status correlates with
marriage. Childless persons were slightly more likely than parents to have at least one
close friend (p < .10).

54

Table A2. Sociodemographic Characteristics by Parental Status and Gender
Characteristic
Entire Sample
(n=1,342)
Among Men
(n=540)
Among Women
(n=802)
Childless
(n=158)
Parents
(n=1,184)
Childless
(n=73)
Fathers
(n=467)
Childless
(n=85)
Mothers
(n=717)
Education (in years) 14.14**** 13.01 14.21 13.46 14.08**** 12.72
Household income (in $1,000) 67.97 73.71 70.04 82.29 66.20 68.12
Multiroles (range 0-3, worker, spouse,
caregiver)
1.77 1.78 1.74 1.91* 1.79 1.70
Married (yes = married or coupled) 0.58 0.77**** 0.56 0.86**** 0.59 0.72***
Caregiver (yes = 1) 0.49* 0.41 0.49* 0.39 0.49 0.43
Employed (yes = 1) 0.70** 0.59 0.68 0.67 0.71*** 0.55
Spouse/partner close (married/coupled) 0.93 0.92 0.95 0.94 0.92 0.90
Have at least one close family member 0.91 0.94 0.89 0.92 0.93 0.95
Have at least one close friend 0.99* 0.97 0.99 0.95 1.00 0.98
*p < .10. **p < .05. ***p < .01. ****p < .001. All ps two-tailed.
Note: Data are means
I used t tests (see Table 3) to compare group differences on caregiving activities
and found that childless people were slightly more likely to be caregivers than were
parents (p < .10). Among parents, 41% were caregivers (490 of 1,184 participants);
among childless people, the percentage was 49% (78 of 158 participants). Men and
women were equally likely to assume the caregiver role; married and single persons had
no difference in their propensity to provide care. Among caregivers, being a woman was
slightly associated with more caregiver burden (p < .10). Marital status and parental
status were not associated with the degree of caregiver burden for caregivers.

55

Table A3. Characteristics of Caregivers by Gender, Marital Status, and Parental Status
Characteristic Proportion being
Caregivers
Gender
Male 40%
Female 44%
Marital status
Married/coupled 42%
Single 45%
Parental status
Parent 41%
Childless 49%*
*p < .10, two-tailed.
Caregiver Burden
Two models of logistic regression were tested to discern the impact of gender,
marital status, and parental status on caregiver burden as expressed by caregivers. The
results indicated that being childless was not related to having greater or lower odds of
having caregiver burden, compared to parents with a close child. However, parents with
no close child were less likely to report having caregiver burden than parents with a close
child (p<.05). Having or not having a resident child did not have a significant effect on
caregiver burden (data not shown). Gender and marital status were not significant factors
in determining caregiver burden (p<.01). As expected, having a close relationship with a
spouse or partner lessened the odds of experiencing caregiver burden (p<.01). Having at
least one close family member other than a spouse or child also helped ameliorate
caregiver burden. Somewhat surprisingly, having at least one close friend increased the
odds of experiencing caregiver burden as expressed by middle-aged caregivers (p<.05).
Table A4 presents the odds ratios of caregiver burden regressed on sociodemographic
56

variables.
Table A4. Main Effects of Logistic Regression of Caregiver Burden on
Sociodemographic Characteristics of caregivers (n = 568)

Variable Model 1
Odds Ratio
Model 2
Odds Ratio
Age 1.002 1.003
Education 0.925** 0.934*
Gender (male = 1) 0.758 0.752
Race (minority=1) 0.578** 0.552**
Childless (= 1) 1.523 1.537
Parent no close child (= 1) 0.504 0.301**
Previously married (= 1) 1.368 1.187
Never married (= 1) 1.182 0.863
Currently employed (= 1) 1.092 1.092
Self-reported health
(problem = 1)
1.455** 1.426*
Household income (in
$1,000)
1.002 1.002
Close spouse/partner 0.426***
Close other family
member(s)
0.487*
Close friend(s) 3.847**
Constant 1.430 1.385
R
2
.047 .082
*p < .10. **p < .05. ***p < .01. All ps two-tailed.
a
Burden = 1.

Depressive Affect
Logistic regression results (see Table A5) indicated that the basic life factors
affecting depressive affect, such as age—the older one got, the lower the odds of having
depressive symptoms (p < .05), and education level, which protected against depressive
57

symptoms (p < .05). As documented in literature, minority participants had greater odds
of experiencing depressive symptoms than Caucasians (p < .01). Not surprisingly, having
a health problem significantly heightens the odds of having depressive affect (p < .001).
Among the predictive factors in this study (i.e., gender, marital status, and
parental status), only marital status showed a significant bearing on depressive affect.
Persons who were separated, divorced, or widowed (the previously married group) had
greater odds of having depressive symptoms than married or coupled persons (p < .01). In
contrast, never-married persons did not have greater odds of having depressive symptoms
compared to married or coupled persons. The caregiver role did not interact with marital
status to affect depressive affect in any way (data not shown).
With regard to parental status and the parent–child relationship, childlessness did
not affect middle-aged persons‘ depressive affect in caregiving situations, nor did the
relationship between parents and children or the existence of a resident child (data not
shown). Caregiving activity by itself did not increase the odds of being depressed for
caregivers, nor did it interact with parental status. Childless persons and parents with no
close child did not fare worse than parents with a close child in caregiving or
noncaregiving situations with regard to depressive affect. I tested an interactive term of
marital status, parental status, and caregiver role, but it did not work because of the small
number of parent with no close child group (n=44).

58

Table A5. Main and Conditional Effects of Depressive Affect Regressed on Parent–Child
Relationship Among Middle-Aged Caregivers and Noncaregivers (n=1,342)

Variable Model 1
Odds Ratio
Model 2
Odds Ratio
Model 3
Odds Ratio
Age 0.966** 0.966** 0.966**
Education 0.942** 0.951** 0.951**
Gender (male = 1) 0.842 0.800 0.798
Race (minority = 1) 1.63*** 1.615*** 1.617***
Childless (= 1) 0.781 0.799 0.797
Parent no close child (= 1) 1.546 1.182 1.457
Previously married (= 1) 2.267**** 2.047**** 2.038****
Never married (= 1) 0.801 0.676 0.690
Caregiver (= 1) 1.217 1.231 1.258
Currently employed (= 1) 0.678*** 0.673*** 0.674***
Self-reported health (problem
= 1)
2.054**** 1.993*** 1.990****
Household income (in $1,000) 0.998* 0.998 0.998
Close spouse/partner 0.519**** 0.518****
Close other family member(s) 0.656 0.652
Close friend(s) 0.660 0.661
Childless × Caregiver 0.998
Parent No Close Child ×
Caregiver
0.573
Constant 4.001 15.016*** 14.707***
R
2
.152 .168 .169
*p < .10. **p < .05. ***p < .01. ****p < .001. All ps two-tailed.
Positive Affect
The results of the logistic regression for positive affect reflected a more
complicated picture than those for depressive affect. Caregiving by itself had no
59

significant impact on a person‘s happiness or enjoyment of life, but it made a difference
for caregivers who had no close relationship with any child. Caregiving had positive
effects on parents with no close child, as their caregiver role interacted with their parent–
child status to increase their odds of being happy or enjoying life (see Models 3 of Table
A6). Although the significance level was just at threshold (p < .10), I regard it worth
reporting as the number of parent caregivers with no close child was as few as 17 in this
sample. I argue that it deserves to be interpreted as significant in view of the low
statistical power from such a small group size. However, among noncaregivers, parents
with no close child had a much lower odds of being happy or enjoying life (p < .01)
compared to parents with a close child. This effect was consistent across all three models
of the regression, as shown in Table A6. The positive effect of caregiving on parents with
no close child should be read together with the lower odds of caregiver burden expressed
by this group of participants (shown in Table A4). As the question identifying caregiver
burden in the HRS questionnaire reads, the lower caregiver burden means that the
caregiving experience is not upsetting, thus, as expected, it does not reduce positive
affect. Having or not having a resident child had no bearing on positive affect (data not
shown; the variable was not included in the final model).
Corresponding to being previously married increasing odds of having depressive
symptoms, it also decreases the chance of having positive affect. Having a close spouse
or partner significantly increased the odds of having positive affect (p < .01). Previously
married persons who were not in couples did not have the benefit of having a close
spouse or partner.
60

As for other sociodemographic variables, age and education level had positive
relationships with positive affect. Having at least a close family member helped (p < .05),
but having a close friend seems not have mattered much.
Table A6. Main and Conditional Effects of Positive Affect Regressed on Parent–Child
Relationship Among Middle-Aged Caregivers and Noncaregivers

Variable Model 1
Odds Ratio
Model 2
Odds Ratio
Model 3
Odds Ratio
Age 1.069*** 1.070*** 1.068***
Education 1.140**** 1.126*** 1.128***
Gender (male = 1) 1.209 1.291 1.327
Race (minority = 1) 0.784 0.804 0.792
Childless (= 1) 0.639 0.617 0.717
Parent no close child (= 1) 0.301*** 0.413** 0.252***
Previously married (= 1) 0.537** 0.617* 0.630*
Never married (= 1) 0.915 1.161 1.034
Caregiver (= 1) 0.951 0.950 0.879
Currently employed (= 1) 1.207 1.220 1.212
Self-reported health (problem
= 1)
0.391**** 0.405**** 0.407****
Household income (in $1,000) 0.998 0.998 0.998
Close spouse/partner 1.913** 1.989***
Close other family member(s) 1.886* 2.004**
Close friend(s) 1.234 1.216
Childless × Caregiver 0.795
Parent No Close Child ×
Caregiver
4.960*
Constant 0.147 0.038* 0.042*
R
2
.107 .123 .129
*p < .10. **p < .05. ***p < .01. ****p < .001. All ps two-tailed.

61

The following figure illustrates the relative odds of having positive affect among
different parental and caregiver groups. Figure 1 compares caregivers and noncaregivers
of different parental statuses. Noncaregiving parents with a close child was the reference
group (odds = 1). Noncaregiving parents with no close child had the lowest odds of being
happy or enjoying life among all parental groups.
Figure A1. Odds ratios of caregivers and noncaregivers of different parental statuses
(reference group is noncaregiving parents with a close child relationship).

*statistically significant

Discussion
Among middle-aged caregivers, do children serve as a mediator for reducing
caregiving stress, or a competitive demand vying for parents‘ resources? Prior research
on the effects of combining parental and caregiver roles on caregivers‘ psychological
well-being does not present a consistent answer to this question. There is no solid
62

evidence for determining how being childless or having a child impacts the psychological
well-being of middle-aged caregivers, an age group more likely than any other to assume
a caregiver role.
Drawing on the data from the lifestyle Leave-Behind Questionnaire of the 2004
HRS, I examined whether and how three different parental statuses (i.e., being childless,
having a close child, and having no close child) affect caregiver burden, depressive
affect, and positive affect among caregivers, in the context of gender and marital status.
The results show that being childless is not related to having greater or lower odds of
having caregiver burden, depressive affect, or positive affect when compared with having
a close child. These findings are applicable to both caregiver and noncaregiver childless
persons. In contrast, parents with no close child benefit from caregiving activities. When
caregiving, parents with no close child experience lower odds of caregiver burden and
increased odds of positive affect for being happy and/or enjoying life. The caregiver role
alleviates the negative effects of being alienated from one‘s own biological children.
Parents with no close child who are not in caregiving roles have the lowest positive affect
compared with childless persons and parents with a close child.
Why does the caregiver role benefit parents with no close child? The result is in
accordance with role enhancement theory, which posits that rewards from one role can
counterbalance demands or stresses from other roles (Sieber, 1974). I speculate that
caregiving may distract caregivers‘ attention from the unsatisfactory relationship with
their child. Assuming a caregiver role may help this group of parents achieve a balance
of the good and bad feelings from their multiple roles, as Voydanoff and Donnelly (1999)
63

found that spending moderate amounts of time in a caregiver role may buffer the effects
of the demand of another role or its distressful effect on psychological well-being.
In this study, gender difference was not detected in the domain of caregiver
burden, depressive affect, or positive affect, at least not reaching statistically significant
level. The slightly greater odds of women having depressive affect but lower odds of
having positive affect than men hinted a possible gender difference, and it was in line
with research findings that women express more depressive symptoms than men in all
age groups throughout their lifetimes.
A surprising finding is that having a close friend actually increases the odds of
having caregiver burden among caregivers (Table A4), whereas having a close
relationship with a spouse or partner and/or having a close family helps with
psychological well-being. This may be explained by the fact that the majority of
caregiving recipients are family members. Therefore, caregivers are more likely to turn to
spouses or partners and other family members for instrumental and emotional support.
This is evidenced by the positive effects of a close relationship with a spouse or partner
and other family members on reducing caregiver burden and depressive affect and
enhancing positive affect. However, caregiving reduces caregivers‘ available social time
with friends, which may contribute to the loss these individuals experience. This finding
is in accordance with the conclusion made by McMullin and Marshall (1996), whose
study found that close family can ameliorate the negative effects of stress on well-being,
but a close friend can not.
Caregiving by itself does not necessarily increase caregiver burden, pose a threat
to positive affect, or increase the odds of having depressive symptoms. Caregiving
64

outcomes have to be considered along with the sociodemographic characteristics of the
caregiver. There was an interactive effect of caregiver role and parent with no close child
status, but not of caregiver role and childless status. It may be interpreted as having no
offspring is not so critical for childless persons, but for parents, the parent–child
relationship matters a lot.
Overall, the relationship with the spouse or partner is more influential than the
relationship with offspring for middle-aged persons with regard to psychological well-
being, as having a close spouse or partner consistently reduces caregiver burden and
depressive affect and enhances positive affect.
Limitations of the Study
Due to data set constraints, it was impossible to know the relationship between the
caregiver and care recipient. The question qualifying caregiving was worded as ―helping
at least one sick, limited, or frail family member or friend on a regular basis.‖ The
ambiguity in the question regarding the care recipient‘s identity makes it impossible to
discern the relationship. Cross-tabulating the caregiver and family structure variables
sheds some light on this issue, though. In my sample, 57.2% of caregivers had at least
one living parent, although only 16.8% of these parents needed assistance with activities
of daily living (ADLs). This is in line with previous research findings that it is more
prevalent for adult children to give instrumental ADL assistance than ADL assistance.
Another potential care recipient is one‘s spouse or child. By cross-tabulating the
caregiver variable and the variable for ongoing physical or emotional problems in the
spouse or child, I found that 53.5% of caregivers confirmed facing such a stressor. The
percentage was higher among parents (55.3%) and lower among childless persons
65

(42.5%). This is understandable, because childless persons have only a spouse‘s health
problem to be concerned about, and not a child‘s. Also, a word of caution is needed here:
Because not all physical or emotional health problems require caregiving, even if a
person indicated facing ongoing physical or emotional problems with a spouse or child,
one cannot automatically assume that this person provided care to his or her spouse or
child frequently. Other possible recipients, such as aging relatives or friends, were not
identified in the dataset either.
It is somewhat unsatisfactory that there is no objective measure of intensity of
caregiving activities in the HRS data set, for example, hours of care or types of care
provided. The general nature of the question (help[ed] at least one sick, limited, or frail
family member or friend on a regular basis) and the reliance on self-classified status of
caregiver may explain why the percentage of caregivers (42%) among this sample
appears to be higher than that (22%) of a similar age group in AARP study (2001).
The Methods section describes two limitations to operationalizing the childlessness
construct. First, as explained previously, because each participant was asked about
number of living children only once upon joining the HRS panel, I had no way of
knowing whether the person suffered the loss of any child(ren) in subsequent waves,
which would have meant that he or she had outlived all children. However, because less
than 1% of the entire HRS sample has outlived all of their children, this should not be a
serious concern. Second, I was unable to discriminate the origin of childlessness—
whether it was by choice or by circumstance. However, the line between voluntary and
involuntary childlessness is often blurred, and adaptation practiced by people through
time may make the distinction less meaningful with regard to outcome. As Cain (2001)
66

observed from her qualitative research on childless women, although some women
actively choose childlessness and some tragically must resign themselves to it, there are a
great many women in between—those who unexpectedly evolve into childlessness by
circumstances, such as late marriage, financial constraints, career development, etc.. The
feelings of women who are childless by circumstance may alter with time or even with
relationships; for example, a woman may regard herself as voluntarily childless if she is
married to a divorced man with children from a previous marriage who does not want to
father any more children even though, if left to her own will, she would prefer to mother
biological children. Therefore, the division between being childless by choice and by
circumstance is often blurred.
The difficulty in classifying childless groups was illustrated by Connidis and
McMulllin (1996). They used a self-identification method to place participants in groups
according to childless by choice or childless by circumstance and asked participants to
state their reason for being childless. The researchers reported that there was a 60%
overlap in the specific reasons given for childlessness among these two groups that
theoretically had distinctly different reasons for being childless. As Dykstra and Hagestad
(2007a) noted, circumstances related to infertility may also be interpreted differently
through time, because perceptions and evaluations are likely to undergo many changes
since the original events.
Although I did not include the variable of having resident child in the final analysis
due to the desire of parsimony in regression models, it deserves some consideration.
Since there is no indication of the ages of participants‘ children, I originally used a
variable of having at least one child at home as a proxy for having underage children,
67

because previous research on combining parental and caregiver roles has focused on
respondents with young children at home. However, there is a possibility that some of the
resident children were actually boomerang children who returned to parents‘ homes after
having already left them once. Theoretically, the age of one‘s children should have a
bearing on caregiving outcomes, as adult children may provide a sense of security
without the stress that accompanies raising young children (Ross & Mirowsky, 2002).
For future research, it may be fruitful to consider children‘s age and independence level,
in addition to parent–child relationship, when assessing caregivers‘ well-being.

68

Chapter Three: Paper B

Childlessness and the Psychological Well-Being of New Retirees
69

Introduction

Retirement is a seminal event in adult life and often brings changes to many
domains of life, such as daily activities, social roles, social contact, and income, just to
name a few. Although most retirees eventually settle into a stable retirement life
(Atchley, 1976) and report satisfaction with life in retirement (Yael, 1984/1985), the
transition process can be stressful, as retirees experience the loss of their work role and
quite possibly a shrinking social network and are faced with uncertainty about the future.
Retirement can be a stressor, and this stress may affect retirees‘ psychological well-being.
The more resources a person has, the less likely he or she will be to suffer life
dissatisfaction caused by retirement-related changes (Pinquart & Schindler, 2007). But
resources are not limited to economic assets; social resources are critical assets, too. As
Pearlin and colleagues (1981) proposed in their stress process model, the resources
available to people can work as buffers against stressors.
For most people, family is the pivotal social resource. Not only does family serve
as a supportive source, but family role—one‘s position in the microsociety of the
family—can work to fill the void left by the lost work role. Although one‘s work role is
often said to be central to a person‘s identity, social roles, as embedded in the network of
family and friends, are more salient in retirement for a worker, compensating for the loss
of the work role and giving a person coherence and continuity in life (Anson,
Antonovsky, Sagy, & Adler, 1989). It is generally believed that the importance of family
relationships grows after a person experiences some type of role loss (Antonucci &
Akiyama, 1987), as is most evidenced with regard to widowhood. This observation is
also applicable to retirement, which signifies a significant role loss for the majority of
70

workers.
Close family members often man the most relied-upon inner circle of one‘s social
convoy and can be strong sources of support in times of life transition such as retirement
(Dorfman, 2002). Research on retirement adjustment has indicated that family is a vital
resource for adjusting to retired life (J. E. Kim & Moen, 2002). In view of the vital role
family plays in retirees‘ adaptation, Szinovacz and Davey (2001) proposed that
retirement and family be studied as interdependent spheres. This makes good sense from
the point of view of the linked life principle of the life course perspective, which
proposes that people‘s life trajectories are shaped by the influences of those close to
them. Apparently, retirement is not just an individual issue, but a family event. How good
a life a retiree has is affected by his or her family.
Although a small body of literature has been established on family relationships
and life satisfaction in retirement, the findings presented in the literature are often
inconsistent. Notably, whether and how retirement impacts a person‘s psychological
well-being remains the subject of much debate (Reitzes, Mutran, & Fernandez, 1996).
Furthermore, how family affects a retiree‘s retirement outcomes in terms of
psychological well-being is not fully understood. With regard to how family members
affect retirement, the importance of the spouse in the retirement decision, transition, and
life satisfaction is more or less consistently established. However, much less is known
about the role adult children and grandchildren play in a retiree‘s retirement outcomes.
Does one‘s offspring, and their offspring, serve as a social resource to mediate the
negative impact and enhance the positive feelings associated with the transition to
retirement? Very few studies have been conducted on this subject. With the intention of
71

filling in this gap in family and retirement research, I investigated the role of children and
grandchildren during the period of the transition to retirement by asking whether the
existence of a child or grandchild functions to ease the transition to retirement, as
manifested in the psychological well-being of the new retiree, regardless of the dynamics
between the retiree and other family members.

Background
Psychological Well-Being in Retirement
In Western society, one‘s work role is central to a person‘s self-identity;
retirement is a major life transition (Anson et al., 1989; J. E. Kim & Moen, 2002).
Although retirement has been recognized as a seminal event in later life, researchers‘
knowledge about its effects on psychological well-being is limited and fragmented (J. E.
Kim & Moen, 2002). The findings from research are often inconsistent and even
contradictory (Szinovacz & Davey, 2006). Uncertainty exists as to whether retirement
has negative, positive, or no effects on psychological well-being (Pinquart & Schindler,
2007). Some studies have found that retirement reduces stress levels (Ekerdt, Bossé, &
LoCastro, 1983; Midanik, Soghikian, Ransom, & Tekawa, 1995) and that retirees enjoy
higher levels of life satisfaction than workers (Isaksson & Johansson, 2000); others have
found extensive disruptive effects for individuals financially, medically, and
psychologically (Covey, 1981; Palmore, Fillenbaum, & George, 1984) or have claimed
that retirement is detrimental to well-being (Hardy & Quadagno, 1995; J. E. Kim &
Moen, 2002; Richardson & Kilty, 1995). Yet another set of studies has shown no
significant impact of retirement on well-being (Calasanti, 1996; Mutran, Reitzes, &
72

Fernandez, 1997; Warr, Butcher, Robertson, & Callinan, 2004) and has argued that the
life satisfaction of retirees remains stable (George & Maddox, 1977).
This lack of consistency in findings is not surprising, considering that studies in
retirement have been done with different designs (some cross-sectional, others
longitudinal), consisted of samples with different characteristics, focused on different
dimensions of social contacts, and used different measures for key variables such as
retirement (Szinovacz & Davey, 2001). All of these findings were accurately drawn from
the respective data according to their respective methodologies, and authors used
competing theories to support their findings. For example, sociological role theory
proposes that employment is a fundamental role central to a person‘s identity (J. E. Kim
& Moen, 2001); thus, the loss of a core role may lead to psychological distress. In
contrast, continuity theory suggests that continuity in retirees‘ other social roles, such as
the family role, may buffer the loss of work role; thus, individuals may maintain their
psychological well-being, and even improve it, through spending time with their family,
engaging in hobbies, enjoying the freedom from work, or volunteering (Wu, Tang, &
Yan, 2005).
The discrepancy in research findings highlights the need for studies that (a) use a
longitudinal data set to tease out the real change before and after the retirement, (b) use a
nationally representative sample to ensure the generalizability of findings, and (c)
measure both positive and negative effects of retirement. Some negative emotions can be
expected, and these can stem from losses associated with withdrawing from the
workforce (e.g., loss of income, reduced social contacts, relinquished work role, etc.;
Anson et al., 1989). However, positive feelings are equally possible, because retirement
73

results in certain gains (e.g., freedom to spend time as one wishes, no hassles of work, a
leisurely life pace, more time with the family, more opportunities to pursue hobbies, etc.).
Influential Factors of Retirement Well-Being
As suggested and empirically supported by Schlossberg (1995) and Fouquereau,
Fernandez, and Mullet (2001), four groups of predictors are significant variables that may
influence adaptation to retirement and thus affect retirees‘ psychological well-being: (a)
sociodemographic characteristics of retirees; (b) resources before and after retirement; (c)
personality and coping responses; and (d) situational variables, such as how the
retirement decision was made. In this study, the variables examined focused on retirees‘
sociodemographic characteristics and family relationships as resources. Some situational
variables were considered, too, but as control variables.
Sociodemographic Characteristics and Psychological Well-Being in Retirement
Age at Retirement
Is it better to retire earlier or later? Research findings are not consistent in this
regard. In a study by Bender and Jivan (2005), retiring at an early age was found to
contribute to retirement satisfaction. However, several other studies have reported that
early retirement is likely to bring less satisfaction than late retirement (e.g., Palmore et
al., 1984; Williamson, Rinehart, & Blank, 1992). The discrepancy in findings may be
better explained by determinants other than age that propel early retirement, such as
health, trade-offs in benefits of work and retirement, and situational factors such as
voluntary or forced retirement.

74

Gender and Retirement Well-Being
Due to the many differences in the genders‘ career trajectories, work histories,
employment opportunities, role conflicts, social norms, and general life experiences,
women adjust somewhat differently than men to retirement (J. E. Kim & Moen, 2002).
Earlier thinking proposed retirement was easier for women, citing women‘s family
orientation (e.g., Barnes & Parry, 2004). For example, married women anticipate more
gains from retirement and have a more positive attitude toward retirement than married
men (Anson et al., 1989). Women‘s decisions about employment are largely influenced
by family demands; for example, mothers in Japan and The Netherlands are less likely
than married women without children to work outside of the home (Koropeckyj-Cox,
2007). Similarly, dissatisfaction or frustration with work may lead some women onto
more domestic, family-oriented life paths (Gerson, 1985). Women used to be more
inclined than men to retire when their spouses did (Moen, 2001), although this is less
common now due to a couple‘s need to maintain resources (e.g., retaining employer-
provided health insurance before becoming eligible for Medicare). In short, there was
once an assumption that women adjusted better to retirement because it is easier for them
to fall back on their family roles of being a housewife or caring for grandchildren.
However, some studies have found that retirement is actually harder for women,
who view it as more disruptive to their lives than men (Atchley, 1976). Women may
perceive more loss in leaving work because they often start or restart their careers later in
life after caring for the family; they do not yet feel ready to retire when the normal
retirement age approaches. Financial uncertainty can be an unwelcome disruption for
women, who often experience more financial constraints in retirement than men due to
75

their precarious careers (J. E. Kim & Moen, 2001), which may lead to lower life
satisfaction in retirement (Seccombe & Lee, 1986). In addition, women are more likely
than men to develop meaningful relationships with coworkers. Retirement may mean
more of a loss in social support for women than for men.
Recent research has found no evidence supporting earlier research findings that
retirement causes more depression for women than for men. The gender differences
observed in retirement research are attributed to the fact that women report slightly lower
levels of life satisfaction and more depression than men throughout life (J. E. Kim &
Moen 2002; Pinquart & Sörensen, 2001). To tease out the real effect of retirement, one
needs longitudinal data with measures taken before and after retirement.
Physical Health
Researchers of retirement have generally agreed that physical health is positively
associated with retirees‘ life satisfaction in retirement (Austrom, Perkins, Damush, &
Hendrie, 2003; Dorfman, 1995; S. Kim & Feldman, 2000; McGoldrick & Cooper, 1994;
Szinovacz & Davey, 2005; Wu et al., 2005). Physical health has proved to be an
important resource in adjusting to retirement. In fact, the influence of health on
retirement well-being occurs long before retirement itself, as workers with bad health are
more likely to be forced to retired, which potentially brings about negative feelings
toward retirement. Poor health can also limit a person‘s pursuit of activities in retirement,
which makes it harder for that person to compensate for the loss of his or her work role
through other social roles or the pursuit of other activities.

76

Socioeconomic Status and Retirement
Education, occupational status, and income are correlated variables, and they
often show positive effects on psychological well-being in retirement (Belgrave & Haug,
1995; Dorfman, 1989; George, Fillenbaum, & Palmore, 1984; Richardson & Kilty,
1991). Having a prestigious occupational status may mean experiencing more loss with
the loss of their work role, but it also gives retirees more resources, such as pensions, for
adapting to retirement and enjoying more leisure and social opportunities (Dorfman,
1989). Highly educated people develop better social skills in coping with changes in life
and can appreciate rewarding opportunities in retirement that enrich their lives (Reitzes &
Mutran, 2004). But not all research agrees that socioeconomic status is relevant to life
satisfaction or psychological well-being in retirement. Some studies have reported no
association at all (S. Kim & Feldman, 2000; Seccombe & Lee, 1986; Wu, Tang, & Yang,
2005). Nevertheless, there is no indication that high socioeconomic status has a negative
association with life satisfaction or psychological well-being in retirement.
Family as a Resource in Retirement
Family members can be a source of instrumental and emotional support for
retirees when the latter need to realign the hierarchy of their personal goals and needs
(Atchley, 1975). In addition, socioemotional selectivity theory suggests that older people
increasingly restrict their social interactions to relationships they find most rewarding—
typically with close family members (Carstensen, 1995). Understandably, this must be
even more the case for older people in retirement. One earlier study showed that there is a
greater chance for retirees to be unhappy and lonely if they do not keep in close touch
with relatives (Murray, 1973).
77

With family structure changing from a pyramid to its present-day beanpole shape
and each generation having fewer within-generation members, intergenerational
relationships become more significant in one‘s life (Giarrusso, Silverstein, & Bengtson,
1996). However, so far, research on retirement effects has often investigated social
contact or social networks in general (Szinovacz & Davey, 2001) and has not paid
enough attention to the influence of younger generations. When studying the family in
retirement, researchers have most often addressed couple relationships, followed by
parent–adult child interactions in terms of familial factors such as geographical distance,
emotional attachment, or filial obligation. Grandchildren are conspicuously missing from
the landscape of retirement research. When researchers have studied the grandparent–
grandchild relationship, they have rarely examined the relationship in the context of
grandparents‘ transitions to retirement (Szinovacz & Davey, 2006). In short, with regard
to family and retirement well-being, most of what researchers know concerns the role of
spouses, followed adult children, and finally grandchildren.
Spouses and Retirement Well-Being
The importance of a spouse in influencing a person‘s retirement decisions,
transition, and satisfaction has been established in the literature (Bender & Jivan, 2005;
N. G. Choi, 1996; J. E. Kim & Moen, 2002). Married persons have advantages over
unmarried persons as evidenced by the fact that married couples have more positive
attitudes toward retirement (Mutran, Reitzes, & Fernandez, 1997) and higher levels of
retirement satisfaction (Price & Joo, 2005). Although most couples desire to retire
together, whether members of a couple retire at the same time is actually not so important
to their psychological well-being (J. E. Kim & Moen, 2002). Some couples even face an
78

unpleasant surprise when they retire at the same time: The strengths and weaknesses of
their marriage may be more closely examined and more acutely experienced in retirement
than before due to the increased amount of time the couple spends together (Anson,
Antonovsky, Sagy, & Alder, 1989).
Never-married elders are considerably less well-off financially than married
couples (N. G. Choi, 1996). Single women, especially those who are divorced or
widowed, are more likely than married women to remain working and retire at an older
age due to financial necessity (N. G. Choi, 2002). Single women are also more likely than
married persons to face financial constraints, evidenced by the fact that Social Security
income accounts for 72% of the formers‘ income in old age, but only 39% of married
persons‘ (Social Security Administration, 2005). Income and wealth are positively related
to retirement satisfaction (Bender & Jivan, 2005). Therefore, there is a belief that married
people adjust to retirement more easily and with more positive outcomes.
Emotional support from a spouse can be very beneficial in reducing the stress
associated with a life transition such as retirement. A satisfactory marriage contributes
considerably to the morale of older men and women (Lee, 1978). In previous research,
the positive effect of marriage on morale was found to be stronger for women than for
men, and more women than men reported that their psychological well-being was more
strongly influenced by the quality of their marriage (Acitelli & Antonucci, 1994; Dehle &
Weiss, 1998; Ross, 1995), regardless of whether they were retired.
Adult Children and Retirement Well-Being
Contrary to the agreement among research findings on the pivotal role of the
spouse in retirement well-being, current research has provided limited information on
79

parent–adult child relationships in the context of employment (Szinovacz & Davey,
2001). Previous studies have shown that retirement impinges on social networks and
one‘s frequency of contact with relatives and friends, but the direction of change is not
clear. Some early studies have suggested that retirement facilitates maintaining close
contact with adult children and grandchildren, but others have failed to support this
finding (Szinovacz & Davey, 2006). Some studies have reported differences between
genders in kin contact and, somewhat surprisingly, that men engage in more kin contact
in retirement and gain more fulfillment than women from contact with adult children
(Szinovacz & Davey, 2001). Overall, the relationship between retirement and offspring
contact is inconclusive; little is known about how offspring affect the psychological well-
being of a person in the transition to retirement.
Grandchildren and Retirement Well-Being
With increased longevity and multiple-generation families becoming more
common, generational ties are becoming more important for older adults (Bengtson,
2001). Increasing life expectancy endows grandparents with more chances to interact
with their grandchildren through the grandchildren‘s childhood and adulthood. In
addition to the longer period of life shared between grandparents and grandchildren,
current grandparenting styles are more relaxed than they were decades ago, and thus the
level of closeness between grandparents and their grandchildren has increased
dramatically. Grandchildren can be a source of joy and fulfillment and provide a sense of
continuity among the generations (Dorfman, 2002). Grandchildren have become an
important source of emotional attachment and social support (Giarrusso et al., 1996), and
80

grandparenthood has emerged as a significant life stage, a big part of which coincides
with retired life.
In view of the fact that 95% of people with children will become grandparents as
well as retirees (Szinovacz, 1998), it is somewhat surprising that there has been such
limited research on retirement and grandparenthood in the past. The majority of research
on grandparenting and later life well-being has focused on the gains and losses of caring
for grandchildren. Little is known about how grandchildren impact grandparents in their
retirement, or whether having a grandchild bears the same meaning and influence for
grandfathers as for grandmothers with regard to psychological well-being in retirement.
Summary of Findings From the Literature Review
In summary, retirement is not a uniform transition, and retirees‘ experiences of
the process may be quite diverse. However, one can see that family plays an important
role in retirement transition outcomes for all, even though data regarding the impact of
children and grandchildren on the psychological well-being of retirees are limited and do
not paint a consistent picture.
The discrepancies concerning retirement well-being found in previous studies are
related to the cross-sectional design of the studies and their failure to control
preretirement status as the baseline against which to assess the real change attributable to
retirement. By not controlling the baseline, research may fall prey to the accumulative
advantage/disadvantage pitfall, resulting in a confounding of the true changes as a result
of the transition to retirement with the effects from preexisting conditions at retirement.
In addition, most existing studies have treated retirement as a single stage from beginning
to end. They have not differentiated the transitional period immediately following
81

retirement from the more settled time further down the road. This can be another reason
for the inconsistent findings.

Research Goals
My first goal in this study was to investigate the impact of offspring on retirement
transition by asking whether a child or a grandchild makes a difference in the
psychological well-being of recent retirees, regardless of the quality or dynamics of the
relationship. As a second goal, I tested whether having a child or a grandchild affects
male and female new retirees differently with regard to psychological well-being.
This was a longitudinal study with pre- and postassessment designed to tease out
the true changes as a result of leaving a work role and assuming a retiree role. Data
analyses compared three different groups of new retirees: (a) childless retirees with no
grandchildren, (b) parent retirees with no grandchildren, and (c) parent retirees with
grandchild(ren). The impact of parental and grandparental statuses was assessed as
isolated from the effects of marital status and gender.

Method
Data Source and Sample Selection
Data Source
Data were drawn from Health and Retirement Study (HRS), which has a
nationally representative household panel surveyed every two years. To achieve a
sufficient number of cases for analysis, I pooled new retirees from the five waves of
survey in year 1996, 1998, 2000, 2002, and 2004. The sample consisted of four sets of
82

retirees from four different time intervals. First set of participants retired between 1996
and 1998, the second set between 1998 and 2000, the third set between 2000 and 2002,
and the fourth set between 2002 and 2004. The year in which a participant was still
working (pre-retirement) was assigned as Time 1 for measurement, and the year in which
a participant reported as retired (post-retirement) was Time 2 for this study. Data files for
each participant contained measures at two consecutive waves—before and after
retirement. The transition period falls within the two-year period, which can vary from
one month to 23 months, depending on how close the retirement occurs before time 2
(post-retirement) measurement.
Sample Selection
To be included in the sample, a respondent must have met the following three
criteria:
1. Working in 1996 but retired in 1998, or working in 1998 but retired in
2000, or working in 2000 but retired in 2002, or working in 2002 but retired in 2004.
2. Reporting as retired for the first time in his or her working life after Time
1. If a person worked at Time 1, retired at Time 2, but reported his or her retirement year
as being before Time 1 (i.e., a series of retirements), then he or she was not selected into
the sample. This was done to exclude those who had retired and had re-entered the job
market, in order to prevent a practice effect of retirement among participants, which can
bias the true change in psychological well-being caused by the new life experience of
transition to retirement.
3. Being between 60 and 74 years old at Time 2. Age 60 was chosen as the
lower threshold because it is close to the average retirement age of 61.5 (National
83

Academy, 2000) for the currently older population and is considered a typical retirement
age. The upper limit was set at 74, because retirement after age 74 is non-normative for
most people. Besides, several studies have found a small decline in life satisfaction
accelerated in the old-old (age 75 and above) due to other loss in life (Schilling, 2006).
Therefore, capping the age limit at 74 reduces the risk of confounding other causes with
retirement effects on psychological well-being.
The final sample in this study consisted of 1,510 respondents aged 60 to 74 who
met the selection criteria above. Originally, there were 1,806 qualified respondents, but
191 among them had missing Center for Epidemiologic Studies Depression Scale (CES-
D) data, which are the outcome variables in this study, and were excluded from the
sample. Another 39 respondents were excluded due to missing data on parental status and
55 childless persons claiming having grandchildren were excluded too since the study
focuses on biological children and grandchildren. Furthermore, 11 participants who had
outlived all of their children were excluded from the pool, resulting in the final sample of
1,510 participants.
Profile of Respondents Excluded Due to Missing CES-D Data
Compared with those Retained in the Sample, respondents with missing
depressive and positive CES-D scores (n=191) were more likely to be male (p < .001), to
have less education (p < .001), to be parents (p < .05), and to be married in retirement (p
< .001). They were also less likely to have held a professional job before retirement (p <
.01). Therefore, the final sample may have better reflected the psychological well-being
of more highly educated professionals and their retirement transition experiences,
particularly for women than for men.
84

Imputation of Missing Data
Several variables had missing data, which is common among longitudinal data
sets. Imputing missing data is one way to prevent a loss of participants, but it poses a risk
of introducing inaccuracy into the final findings. Fortunately, only three variables needed
imputation, and almost all of them had less than 10% missing data. The hot deck
imputation method was used to impute missing values. Those four imputed variables
were (a) forced to retire (1 = forced to retire; missing 168, 10.7%), (b) professional job
before retirement (1 = holding professional job; missing 138, 8.8%), and (c) household
income (missing 140, 8.9%).
Measures
Definition and Operationalization of Retirement
There is no uniform definition of retirement (Gustman, Mitchell, & Steinmeier,
1995). For some, it may mean complete withdrawal from work force and starting to
collect a pension (Richardson, 1999). For others, it may mean simply a reduction in
working hours from full time to part time. Part-time workers can further be divided
between those who have always worked part time and those who previously held full-
time employment but who now work part time. Or one can still work, but not for
monetary reward, as in the case of volunteers. These are just some examples of the array
of objective and subjective definitions of retirement. Ambiguity in the definition of
retirement may result in distorted findings in research. Even with the same data set, using
different measures for retirement can lead to different tallies for employment and
retirement statuses. Taking the data of HRS as an example, researchers found that fewer
85

people in that data set subjectively defined themselves as retired than worked for zero or
few hours per year (Gustman, Mitchell, & Steinmeier, 1995).
This study adopted the self-defined status as the indication of retirement. Whether
a respondent was retired was determined by the status that participant assigned to himself
or herself when responding to the following question: ―Are you working now,
temporarily laid off, unemployed and looking for work, disabled and unable to work,
retired, a homemaker, or what?‖ If a respondent chose any other status than ―retired‖ at
Time 1, but subsequently answered ―retired‖ at Time 2, he or she was considered retired.
A separate question inquired about the year in which a person retired. If the year given
was not earlier than Time 1, the person was assigned a status as a new retiree and was
included in the sample.
Approximately 1% of respondents assigned their employment status as ―others‖
or ―don‘t know‖ or ―refuse to answer.‖ They were excluded from the sample pool.
Definition and Operationalization of Childlessness and grandchildlessness
In this study, childlessness is defined as the status of never having had a
biological child, regardless whether one has an adopted child or stepchild. This is the
most widely adopted definition of childlessness in aging research (Koropeckyj-Cox,
1998). I chose to limit the definition of childlessness to biological children because of
data constraints in the HRS that do not provide information on adopted children, and also
because of the inherited uncertainty associated with the stepparent–stepchild relationship.
The HRS asks respondents about their parental status when they are first
surveyed. The question is not put to them in the successive waves because the majority of
participants are aged 50 and older (unless that respondent is a much younger spouse of
86

the primary respondent) and are assumed to have passed the age of childbearing. The
question is worded as follows: ―How many children have you fathered/have you given
birth to? Please don‘t count miscarriages or still-births, or adopted or step-children for
this question.‖
The question regarding whether a respondent has ever had a biological child is
followed by one that surveys the number of living children: ―How many of them are still
living?‖ As with the question about having had a biological child, the question about the
number of living children is asked the first time a participant joins the panel but not again
in successive waves. This poses a potential problem of being unable to identify
respondents who have lost all of their children after the first wave of participation in the
HRS. However, the number of parents who outlive their children is small (less than 1%
of the entire HRS sample); therefore, this limitation should not impact my study
significantly.
To compile a complete set of parental status information, I compared the number
of biological children and the number of living children. If the number of biological
children was at least 1, and the number of living children was not 0, then the person did
not meet the criterion of outliving all children, and I recorded the value of outliving-all-
children variable as 0. If the person did not have a biological child, there was no question
about surviving all children, so the outliving-all-children value was also 0. Only when a
person had biological child(ren) and no living child was he or she counted as having
outlived all children and the value for that variable entered as 1. A person who survived
all children was excluded from the sample for this study.
87

Grandchildlessness is defined as having no biological grandchild. In HRS, the
number of grandchildren question is asked as follows: ―Altogether, how many
grandchildren do you (and your (husband/wife/partner) have?‖ If the answer is zero,
then, the person is regarded as having no grandchildren. Note that this question actually
does not differentiate biological grandchild from step-grandchild or adopted-grandchild.
It is easy to identify in the case of childless persons. If a childless person claims to have
grandchild(ren), that cannot be a biological relationship, thus I excluded him or her from
the sample pool. However, it is harder to differentiate in the case of parents. It is assumed
that at least one of their grandchildren was born to their biological child thus they do have
at least one biological grandchild. However, this does not fully cancel out the possibility
that the grandchild is adopted by their child or is a step-child to their child.
In this study, participants are classified into three groups: 1) childless retirees with
no grandchild, 2) parent retirees with no grandchild, and 3) parent retirees with at least
one grandchild. Parent retirees with no grandchild form the reference group, which
childless retirees and grandparent retirees are compared against to detect the effect of
having children and having grandchildren respectively.
Definition of Psychological Well-Being
Analogous to the idea that health is not merely the absence of disease,
psychological well-being is not merely the absence of depression. Psychological well-
being is a multidimensional concept that acknowledges the potential coexistence of both
positive and negative feelings. Therefore, when assessing psychological well-being,
researchers should investigate the manifestation of both positive and negative aspects of
emotions.
88

Psychological well-being is measured in the HRS with the CES-D, a self-report
depression scale for research in the general public, which was designed to measure
depressive symptoms by emphasizing their subjective and affective elements (McDowell
& Newell, 1996; Radloff, 1977). The CES-D has been widely used in research on the
general and clinical populations (Radloff, 1977), and it has good internal consistency
(Hann, Winter, & Jacobsen, 1999). Three indicators of psychological well-being—
depression, loneliness, and sadness—composed the measure of depressive affect in my
study. They are traditionally cited in normative admonitions against childlessness
(Koropeckyj-Cox, 2002). Loneliness has been defined as an unpleasant feeling of
dissatisfaction with either the number or the quality of existing social relationships
(DeJong-Gierveld, 1987; Koropeckyj-Cox, 1998, 2002; Perlman & Peplau, 1982).
Depression is ambiguous in definition and has various manifestations of feeling low, blue
or simply down (Blazer, 1995). The manifestation of positive aspects of well-being is
recognized in the form of global happiness, satisfaction with family life, and satisfaction
with friendship in most adopted measures. In the CES-D (Radloff, 1977), they are
measured as ―being happy‖ and ―enjoying life.‖
The HRS adopted a 9-item battery of CES-D items, with each item limited to a
dichotomous yes/no response. Six of the nine items indicate the presence of depression as
manifested in negative feelings and somatic symptoms, and three indicate positive
feelings. Positive feelings include being happy, enjoying life, and having a lot of energy
much of the time during the past week; negative feelings include being depressed, lonely,
or sad much of time during the past week; somatic indicators include feeling that
everything is effort, sleeping restlessly, and finding it hard to get going. The lead-in
89

question for the CES-D in the HRS reads as follows: ―Now think about the past week and
the feelings you have experienced. Please tell me if each of the following was true for
you much of the time during the past week. Much of the time during the past week …‖
Several published papers have used the CES-D to study mental health based on HRS data
(Gallo, Bradley, Siegel, &Kasl, 2000; Siegel, Bradley, Gallo, & Kasl, 2003, 2004), and
Gallo and colleagues concluded that it has acceptable measurement properties after
assessing its reliability and validity using 1992 and 1994 HRS data.
In this study, individual CES-D items were grouped into two ordinal indexes:
depressive affect and positive affect of psychological well-being. The following steps
were involved in the transformation: (1) using factor analysis to extract a positive affect
factor and a depressive affect factor from the nine items, and (2) adding together items
belonging to the same factor to form an ordinal index. The factor analyses extracted the
two factors: (a) depressive affect factor, with three original CES-D items of being
depressed, feeling lonely, and feeling sad, and has a value ranging from 0 to 3, with 0
means no depressive symptoms, and 3 stands for three depressive symptoms (the
maximum in this study); and (b) positive affect factor, with two original CES-D item of
being happy and enjoying life. Although I initially transformed the original discrete
dichotomous measure into an ordinal measure, I treated depressive affect and positive
affect as a dichotomy for the logistic regression analysis. In the regression, depressive
affect was either 0 (no depressive symptoms at all) or 1 (one to three depressive
symptoms). Positive affect was coded as 0 (not being happy and not enjoying life) or 1
(being happy or enjoying life, or both).
90

The 3-item depressive affect factor had reliability scores of .664 and .706 for pre-
retirement and post-retirement, respectively. The 2-item positive affect factor had a
reliability of .597 for preretirement and .657 for postretirement.
Predictive Variables
The key predictive variables are the combination of parental and grandparental
status—whether a person is childless and grandchildless, or have child but no grandchild,
or have both child and grandchild. Other predictors are marital status and gender. Marital
status includes two categories: being married or being unmarried in retirement.
Control Variables
Controlled variables included important demographic characteristics and other
factors that are evidenced to be associated with retirement satisfaction, such as age of
retirement, educational level, household income, and health in retirement (Bender &
Jivan, 2005). The variable of health status in the HRS is measured with a self-reported
item with a scale of 1 to 5 (1 = excellent, 2=very good, 3=good, 4=fair, and 5 = poor).
The situational factors associated with retirement controlled in this study are occupational
status (professional job=1, which include managerial and professional job), and being
forced to retire or not (being forced=1).
Initially, I include a variable of ―child living within 10 miles‖ in the analysis,
which turned out to be highly correlated (.799) with the status of ―parents have
grandchild‖ (but not significantly correlated with ―parents have no grandchild‖). The high
correlation skewed the regression results. Therefore, I excluded the variable of ―child
living within 10 miles‖ from final analysis.

91

Results
Profile of Participants
Respondents in this sample were born between 1929 and 1944. Many of them
were parents of baby boomers. It was not surprising that the childless rate in this group
was only 7.9%, which was lower than that of both the generation before (the Great
Depression generation) and the generation after (the baby boom generation). After I
excluded 55 childless persons who claimed to have grandchildren, the childless
population with no grandchildren made up only 4.7% (n = 71) of the sample. Parents with
no grandchildren accounted for 9.7% (n = 147) of the sample, whereas parents with
grandchildren composed the majority of the sample (85.6%; n = 1,292).
In this study, men and women were represented in approximately equal numbers.
Two thirds were married and retired. Approximately one third of respondents had retired
from a professional job, and a quarter of the sample felt that they had been forced or
partially forced to retire. Table B1 provides a sketch of sample characteristics for this
study.
92

Table B1. Demographic Characteristics of Respondents (n = 1,510)
Characteristic n %
Gender
Male 743 49.2
Female 767 50.8
Race
Caucasian 1,257 83.2
Non-Caucasian 253 16.8
Marital status in retirement
Married 1,030 68.2
Not married 480 31.8
Parental/grandparental status
Childless no grandchild 71 4.7
Parent no grandchild 147 9.7
Parent with grandchild 1,292 85.6
Child within 10 miles
Yes 930 59.7
No 635 40.3
Professional job before
Yes 471 31.2
No 1,039 68.8
Forced retirement
Yes 389 25.8
No 964 74.2
Volunteer in retirement
Yes 546 36.2
No 964 63.8

Descriptive analysis showed that the participants had a mean educational level of
12.71 years and an average retirement age of 64.03. In general, respondents reported their
health to be worse in postretirement than in preretirement; self-rated health score had a
mean of 2.48 preretirement and 2.61 postretirement, where 1 = excellent and 5 = poor.
The sample as a whole suffered few depressive symptoms (M = 0.40 out of a total
of 3 depressive symptoms in preretirement, and M = 0.42 in postretirement); they also
enjoyed a high positive affect score (M = 1.86 out of a total of 2 in preretirement and M =
1.85 in postretirement).
93

However, members of different groups did not suffer depressive affect or enjoy
positive affect equally. Table B2 compares the mean level of psychological well-being of
group members with different genders, marital statuses, and parental and grandparental
statuses. It shows that women consistently reported more depressive symptoms than men
before and after retirement (ps < .01) and were slightly less likely to be happy or enjoy
life postretirement (p < .10). But there was no difference in the level of net change in
psychological well-being before and after retirement between genders. As expected,
unmarried persons showed significantly higher depressive levels and lower positive
levels than married persons before and after retirement (ps < .01). Again, there was no
difference in the level of net change in psychological well-being before and after
retirement between the married and the unmarried groups.
On the front of parental status, at first glance it seems that childless persons were
more depressed than parents and also less happy or enjoyed life less in retirement.
However, this may not be an accurate conclusion, as the childless group had a greater
proportion of unmarried persons than the parent group (61% vs. 30%, respectively), and
being unmarried is highly associated with higher depressive affect and lower positive
affect, as discussed previously. A further analysis with logistic regression teased out the
effects of parental status from marital status (see Tables B3 and B4).
It is interesting that having grandchildren became significant after retirement, as
there was no association of grandparenthood with psychological well-being until
postretirement. The difference associated with grandparenthood was seen in positive
affect but not in depressive affect.
94

Table B2 Mean Scores for Depressive and Positive Affect

Characteristic Depressive Affect
(Range 0–3)
Positive Affect
(Range 0–2)
Pre-
retirement
Post-
retirement
Pre-
retirement
Post-
retirement
Gender
Male 0.33 0.34 1.86 1.88
Female 0.47 0.51*** 1.85 1.83*
Marital status
Married 0.30 0.30 1.89 1.89
Unmarried 0.61*** 0.69*** 1.79*** 1.77***
Parental/grandparental
status

Childless no grandchild 0.59 0.61 1.80 1.75
Parent no grandchild 0.42 0.40 1.86 1.80
Parent with grandchild 0.39 0.42 1.86 1.87*

*p < .10; ***p < .001.

Effects of Children and Grandchildren on Retirement Well-Being
I used logistic regression to identify the significance of the predictive variables. I
combined parental status and grandparental status to classify participants into three
groups: childless persons with no grandchildren, parents with no grandchildren, and
parents with grandchildren. Parents with no grandchildren served as reference in the
regression.
On Depressive Affect
Findings indicated that, for the entire sample, having a biological child was not
related to depressive affect. Having a grandchild did not influence depressive affect
either. What determined the level of depressive symptoms in retirement were (a) being
married in retirement, which significantly reduced the odds of having depressive affect
(p < .001); (b) self-rated health in retirement (the better one‘s health, the lower the odds
95

of suffering from depressive affect; p < .001); and (c) conditions of retirement (if a
person had been forced to retire, he or she suffered much higher depressive affect in the
transition to retirement; see Table B3). Preretirement depressive affect was significant in
determining the level of postretirement depressive affect during the transition period.
Table B3 Effects of Parental and Grandparental Status on Depressive Affect in the
Transition to Retirement (n= 1,510)

Variable Depressive Affect
Model 1 Model 2
Gender (ref = male) 0.868 0.877
Education 0.997 0.996
Race (ref = minority) 1.019 0.997
Age at retirement 1.016 1.007
Martial status in retirement (ref =
unmarried)
0.499**** 0.505****
Parental/grandparental status (ref =
parent no grandchild)

Childless no grandchild 1.079 1.194
Parent with grandchild 0.996 1.027
Self-rated health in retirement (1 =
excellent, 5 = poor)
1.394**** 1.326****
Household income (in $1,000) 0.999 0.999
Preretirement depressive affect (range =
0–3 symptoms)
2.482**** 2.480****
Professional job before retired (ref =
yes)
1.167
Forced retirement (ref = forced) 2.095****
Volunteer in retirement (ref = yes) 0.986
R
2
.240 .262
Note: Data are odds ratios.
****p < .001.
On Positive Affect

96

In the domain of positive affect, the predictors for a retiree to transition to
retirement with positive feelings are enjoying good health (p < .001), feeling that
retirement was not forced (p < .001), enjoying life and being happy prior to retirement (p
< .001), and having a grandchild (p<.05).
Childless persons were not significantly different from parents with no
grandchildren in positive affect, but parents with grandchildren enjoyed a higher odds of
having positive affect. Apparently, grandchildren bring joy to grandparents‘ life and
make them happy.
Having no biological children implies a disadvantage for positive affect when one
considers that having no biological children leads to having no biological grandchildren,
thus resulting in having no chance to enjoy the benefits of having a biological grandchild.
A separate analysis (data not shown) comparing the childless group directly with parents
with grandchildren showed that childless persons had slightly lower positive affect than
parents with grandchildren (p < .10).

97

Table B4 Effects of Parental and Grandparental Status on Positive Affect in Transition to
Retirement (n=1,510)

Variable Positive Affect
Model 1

Model 2

Gender (ref = male) 1.119 1.120
Education 1.060 1.069
Race (ref = minority) 0.961 1.004
Age at retirement 0.939 0.939
Marital status in retirement (ref =
unmarried)
1.600 1.563
Parental/grandparental status (ref =
parent no grandchild)

Childless no grandchild 1.533 1.327
Parent with grandchild 2.648** 2.567**
Self-rated health in retirement (1 =
excellent, 5 = poor)
0.578**** 0.618****
Household income (in $1,000) 1.002 1.002
Preretirement depressive affect (range =
0–3 symptoms)
2.827**** 2.700****
Professional job before retired (ref =
yes)
0.630
Forced retirement (ref = forced) 0.455***
Volunteer in retirement (ref = yes) 1.284
R
2
.168 .189
Note: Data are odds ratios.
**p < .05; ***p < .01; ****p < .001.

Gender Differences in the Transition to Retirement

Several differences between men and women jumped out from the regression of
psychological well-being on sociodemographic variables (see Table 5). It was obvious
that men were more prone to the negative impact of delayed retirement. Retiring at a late
age generally decreased positive affect for men (p < .05), regardless of the conditions of
retirement (i.e., voluntary or forced retirement). But women were not subject to the effect
98

of retiring early or late. Marriage has more power for men than women when in the
transition to retirement, as married men were less likely than unmarried men to be
depressed (p < .001) but were more likely to be happy and/or enjoy life (p < .05). In
contrast, married women were not necessarily more likely to be happy or enjoy life,
although they were also less likely to be depressed than unmarried women (p < .05).
Childless men had lower odds of depressive affect, whereas childless women
were more susceptible to depressive affect, although this did not reach statistically
significant levels in either case. In contrast, having a grandchild had a positive effect for
grandfathers transitioning to retirement, manifesting in the grandfathers‘ positive affect,
as grandfathers had greater odds than fathers without grandchildren of being happy and
enjoying life (p < .05). But grandmothers did not have an advantage over mothers without
grandchildren with regard to positive affect.

99

Table B5. Variables Affecting the Psychological Well-Being of Men and Women
Variable Men’s Affect

Women’s Affect
Depressive

Positive

Depressiv
e

Positive

Education 0.981 1.049 1.029 1.070
Race (ref = minority) 1.195 1.328 0.862 0.908
Age at retirement 1.046 0.857** 0.977 1.022
Marital status in retirement (ref
=unmarried)
0.398**** 2.712** 0.610** 1.020
Parental/grandparental status (ref =
parent no grandchild)

Childless no grandchild 0.806 2.585 2.049 0.791
Parent with grandchild 0.885 3.472** 1.324 1.924
Self-rated health (1 = excellent, 5 =
poor)
1.411*** 0.628** 1.271*** 0.628***
Household income (in $1,000) 1.001 1.002 0.996 1.005
Preretirement depressive affect
(range = 0–3 symptoms)
2.792**** 3.504**** 2.292**** 2.071**
Professional job before retired (ref =
yes)
1.412 0.931 0.983 0.618
Forced retirement (ref = forced) 2.632**** 0.442* 1.774*** 0.482*
Volunteer in retirement (ref = yes) 0.809 0.858 1.108 1.604
R
2
.291 .276 .243 .151
Note: Data are odds ratios.
*p < .1; **p < .05; ***p < .01; ****p < .001.

Discussion and Limitations

In this study, I investigated whether the existence of a child or a grandchild makes
a difference in the psychological well-being of older adults in the transition to the retiree
role, and whether the effects are different for men and women. The purpose was to
understand whether children and grandchildren serve as resources for helping new
retirees move into the retirement stage, as this transition can be stressful due to changes
100

affecting many domains of life when the work role is no longer part of a person‘s
identity.
Generally speaking, the new retirees in this sample enjoyed high positive feelings
and had low depressive symptoms. Some differences between groups did exist, with
married people having significantly less depressive symptoms in both preretirement and
postretirement compared with divorced, widowed, and never-married persons. As
demonstrated in the literature, a spouse is the primary support source for a person in
stressful events. However, being married in retirement is more beneficial for men than for
women. Compared with unmarried men, being married reduces married men‘s odds of
having depressive symptoms and increases their chance of being happy and enjoying life.
Compared with unmarried women, being married only reduces married women‘s odds of
having depressive symptoms; it does not increase their chance of being happy and
enjoying life. Furthermore, the effect on reducing relative odds of depressive affect is
larger for married men than for married women, when they are compared with their
respective unmarried counterparts.
It appears that children are not as effective as grandchildren in enhancing the
psychological well-being of new retirees. Having or not a child is not a determinant of
psychological well-being for new retirees, as no difference in relative odds, either in
depressive or positive affect, was detected when childless persons with no grandchildren
were compared with parents with no grandchildren. However, the lower relative odds of
childless men having depressive affect compared to fathers, in contrast to the higher
relative odds of childless women compared to mothers, are suggestive of the possibility
that parenthood means different for men and women, even although they were not
101

statistically significant. This seems to hint that motherhood is more salient for women‘s
psychological well-being than fatherhood is for men in old age, as suggested by previous
research.
However, having a grandchild did make difference as evidenced in the contrast
between grandparents and parents with no grandchildren. The existence of a grandchild
enhances positive affect for new retirees, making retired grandparents happier and/or
helping them to enjoy life more. This effect actually lies with grandfathers more so than
with grandmothers. When grandfathers were compared with fathers without
grandchildren, they had a higher odds of being happy and/or enjoying life, but
grandmothers did not have this same advantage over mothers without grandchildren, at
least not one that reached a statistically significant level. The positive impact of
grandparenthood was detected in positive affect but not in depressive affect.
The finding that having grandchildren is beneficial for positive affect but not for
depressive affect is particularly interesting because it stands in sharp contrast to being
married in retirement, which significantly reduces depressive affect but does not enhance
positive affect (in women‘s case). It appears that spouses are a reliable source of support,
whereas grandchildren provide pure enjoyment and happiness in life. This finding also
supports the argument that, when measuring psychological well-being, one should
consider both positive and negative spheres to capture a fuller range of emotions.
This study confirms that family is an important resource for new retirees for
maintaining psychological well-being when transitioning from the work role to the retiree
role. The findings help to eliminate, to some extent, the confusion arising from
inconsistent findings of previous studies on whether family really matters, and which
102

family relationship help, and in what way, for the retirement transition. This study
demonstrates the basic, fundamental value of family. The existence of a spouse reduces
depressive symptoms for both male and female retirees and increases positive affect for
male retirees; the presence of a grandchild is effective in enhancing positive feelings for
grandfathers. The study affirms family as a resource in the retirement transition and
shows that family roles compensate for the loss of the work role, even more so for men
than for women.
Combining the findings that having a grandchild is positive for grandfathers, and
that marriage holds greater power for men than for women in the retirement transition,
one can see that the work role has been central for older men due to their socialization. It
can be inferred that retirement may create a bigger psychological void for men than for
women. It appears that men take more from their family role as a spouse and as a
grandparent to compensate for the loss of their work role. These findings also conform to
the known fact that men rely on their spouses for emotional support, whereas women
may have more potential resources from a larger social network. Therefore, although
women lose their work role in the transition to retirement, they still have other sources of
social support than their family to help them transition, which reduces their reliance on
their role as spouse and grandmother.
Not only does family role have different levels of importance for men and
women, but age at retirement has a different impact as well. Retiring at a later age was
negatively related to positive affect among men but not so among women. I speculate that
men normally have a longer working history than women. Therefore, they may welcome
the chance to retire early. But the prospect of early retirement may not be as attractive for
103

women because many often start or reenter the job market later after raising their
children, and they may want to continue longer in their work life.
One point worth noting from the regression table (Table 5) is that the same set of
sociodemographic variables explains much less of women‘s positive affect than men‘s
(15.1% vs. 27.6%, respectively). I speculate that more complicated contextual factors
associated with women‘s retirement decisions and timing (e.g., the need to provide care
or meet the expectation of retired spouses) affect how happy women feel or how much
they enjoy life immediately after retirement. This confirms that a retirement model for
men may not be applicable to women, as many retirement researchers (e.g., J. E. Kim &
Moen, 2002) have pointed out in the past decade.
With the aging of the population and baby boomers on the verge of retirement,
today‘s workers are retiring in record numbers . Their experiences during the transition to
retirement will be felt by society as a whole. It is important to understand which factors
enable new retirees to enjoy the positive outcomes of retiring. This study contributes to
the understanding of the impact of family in the transition to retirement. The findings
confirm the impact of spouses, explore the function of grandchildren, and shed light on
gender differences in the retirement process.
One particularly notable finding is that the grandparent–grandchild relationship
emerged as an important factor contributing to psychological well-being for new retirees,
and particularly for grandfathers. With longevity and the beanpole family structure
prevalent in modern time, the importance of grandparenthood is more salient than ever.
However, the almost 20% childless rate among baby boomers will reduce the number of
future grandparents, as childless persons will not have grandchildren.
104

How does society make the benefits of grandparenthood available to future older
adults, even if they do not have a biological grandchild? The findings from this study
have policy implications for encouraging programs that facilitate intergenerational
closeness among biological or nonbiological grandparents and grandchildren. This will
be particularly important for the baby boom generation, because it is predicted that 25%
of this generation will enter old age without a spouse or a child (Wachter, 1997). Related
to this is the fact that, with blended families becoming the most common family type in
America, the relationship between step-grandparents and step-grandchildren is worth
further research.
Limitations of the Study
Approximately 10% of the respondents in the sample were couples. The inclusion
of spouses from the same couple in the analysis rendered the sample not strictly
independent and may have biased the findings with potential statistical interdependence
problems, although I do not see this as having a significant effect on the results.
Another limitation was the variation in the length of time between retirement and
the postretirement measure, which ranged between 1 and 23 months among participants.
If a participant had just retired shortly before the Time 2 measurement, then he or she
was still in the very early stages of postretirement. However, if a participant had retired
immediately after the Time 1 (preretirement) measure and was surveyed again 2 years
later in the subsequent wave of the HRS, then 23 months had elapsed. This posed a risk
of treating all new retirees as if they were in the same point in retirement process,
although the former group was still in honeymoon phase and the later group had
advanced to the realization or stabilization periods, to use Atchley‘s (1974) terminology.
105

Nonetheless, not all retirees experience these stages in an orderly or consistent fashion
anyway (Atchley, 1974). This is an unfortunate limitation, but it should not significantly
affect the findings of a representative sample such as that used in the Health and
Retirement Study.

106

Chapter Four: Paper C
Childlessness, Care Provision, and the Psychological Well-Being
of Older Adults With Disabilities
107

Introduction
Spouses are the preferred helpers among older adults in need of assistance to
perform activities of daily living (ADLs), followed by adult children (Messeri,
Silverstein, & Litwak, 1993). Although adult children are the second line of defense for
older persons fending off difficulties associated with old age (Rossi & Rossi, 1990), they
far outnumber spouses when it comes to being the active caregiver (Spillman & Pezzin,
2000) due to the high rate of widowhood among the older population and the fact that
siblings often share responsibility for caring for their parents. Adult children play a large
role in providing aid to parents faced with widowhood and needing care in advanced age.
Research has found that widowed parents who have disabilities receive more support than
their childless counterparts (Boaz & Hu, 1997; N. G. Choi, 1994; Gironda, Lubben, &
Atchison, 1999; Johnson & Troll, 1992; Larsson & Silverstein, 2004). Although childless
persons have more active ties with siblings, friends, and other relatives than parents do,
there is little evidence that having active ties guarantees receiving instrumental support
when it is needed (Chappell & Badger, 1989; Lee & Ihinger-Tallman, 1980). Similarly,
unmarried and childless persons tend to confide in, and rely on, distant kin and friends for
companionship. Nonetheless, these relationships, although vital in meeting the need for
sociability, are not necessarily a good source of long-term ADL assistance (Johnson &
Catalano, 1981).
With the shrinking of the caregiver pool due to societal changes such as women‘s
labor force participation and family structure changes, current trends in long-term care
signify more reliance on formal care than ever before. The increase in formal care usage
complements, but has not replaced, informal care use (Chappell & Blandford, 1991;
108

Patsios & Davey, 2005). Family members, particularly adult children, often arrange
formal care for older adults in need via the engagement of care managers. The formal
care repertoire often includes assistive devices recommended by care managers.
Assistance devices, defined as ―any device that is designed, made, or adapted to assist a
person perform a particular task‖ (MedicineNet.com, 2003), have been hailed by policy
makers as a venue to reduce the intensity of personal care, to enhance the independence
of persons with disabilities, and to curb long-term care costs (Burdick, 2007). Assistive
devices are increasingly important in the repertoire of care provision and often coexist
with personal assistance from formal or informal caregivers as means of alleviating the
functional limitations of persons with disabilities.
A person who has limitations in functional ADLs, such as walking across a room,
getting into and out of bed, eating, bathing, dressing, or toileting, is classified as being
disabled (World Health Organization, 1993). Disability is a stressor in life as it prevents a
person from conducting his or her life in a typical way. Social support and instrumental
support are buffers to stress (Bowling, 1994, 2005; Lakey & Cohen, 2000). How does the
absence of children—the purported primary source of instrumental support—impact
aging persons with disabilities? Do the trends toward less informal and more formal care
among the older population make children less crucial to care availability? Or do
childless people indeed receive less help compared to parents when faced with disability?
If so, how does this reflect in their psychological well-being? So far, little is known in the
domain of childlessness and disability.
For this study, I investigated how parental status, within the context of gender and
marital status, affects ADL care provision among the old-old (aged 75 and older). The
109

research goals were to understand (a) the effects of childlessness on assistance with
ADLs, (b) the entailed outcomes of the provision of different ADL assistance on
psychological well-being, and (c) the direct effects of childlessness on the psychological
well-being of older adults with disabilities in the context of gender and marital status.
The purpose of this study was to identify the vulnerable group for long-term care policy
consideration.
Literature Review
Family as a Support Source
A significant body of literature exists that describes and explains the support
structures of older adults. A good portion of these studies have investigated instrumental
support for ADL difficulty. In studying ADL support structure, researchers have often
adopted the framework of the hierarchical–compensatory model or the task-specific
model. Two prominent models describe and predict the support structure for older adults
(Messeri, Silverstein, & Litwak, 1993). The hierarchical–compensatory model proposes
that elders prefer and choose their caregivers according to the primacy of relationship, in
the order of spouse, children, relatives, friends, neighbors, and, finally, formal
organizations (Cantor, 1979; Shanas, 1979). The task-specific model, a response to
dissatisfaction that the hierarchical–compensatory model does not fully explain research
data, suggests that there is no global sequence of support providers; instead, elders
choose their support sources according to the degree of compatibility between providers‘
properties and task characteristics. Providers‘ properties are classified into seven
dimensions, namely proximity to the person in need, length of commitment, common
lifestyle, size, sources of motivation, division of labor, and level of technical knowledge
110

needed to perform the task (Litwak, 1985). It is important to note that, within the domain
of ADL, both the hierarchical–compensatory model and the task-specific model predict
older adults will engage support sources in the same order (i.e., spouse as the primary
choice, followed by adult children; Messeri et al., 1993).
It has long been established that spouses and adult children are the primary
sources of ADL assistance (S. Allen, Goldscheider, & Ciambrone, 1999; Dykstra, 1990;
Wellman, 1992). Persons without spouses may suffer a more fragmented and potentially
less stable arrangement of care options (Freedman, Aykan, Wolf, & Marcotte, 2004), and
it is not clear whether adult children can fill the void left by a lost spouse. Individuals
without spouses and without children may suffer further reduced resources. Although
never-married women build up substitute support networks throughout life, how effective
their networks are for ADL support is not clear.
In short, spouses and adult children are the primary providers for ADL support,
and they are not easily substituted with other relationships. Formal care is an increasingly
common component of care received by older adults with disabilities, and it is often
arranged by family members, particularly adult children. From the previous discussion, it
is clear that to understand the nature of care provision among older adults with
disabilities, one has to look at the impact of marital status and parental status as well as
the possible interaction between them.
Marital Status and ADL Support
Divorced Persons
Older divorcees are vulnerable to having limited informal support (Connidis &
McMullin, 1994). This is even more true for divorced fathers than divorced mothers. As
111

sociologists note, women are the kinkeepers of families. When severing marriage ties
from their spouses, divorced fathers lose not only spousal support, but possibly also their
connection with children and other social ties, resulting in their having the smallest social
networks and very limited family contact (Mugford & Kendig, 1986). Generally
speaking, divorced women have larger networks than divorced men and rely on friends
for emotional support (Gibson & Mugford, 1986). However, emotional support has
different requirements than instrumental support; whether emotional support can translate
to ADL support for divorced women is questionable.
Widowed Persons
Connidis and McMullin (1994) reported that widowhood increases emotional and
financial support for widowed persons, but not instrumental support such as assistance
with ADLs. This discrepancy in type of support can be explained by the substitution
principles of the task-specific model. Neither emotional support nor financial support is
necessarily as demanding as instrumental support, and in particular ADL support.
Because few relationships possess the same properties as the spousal relationship, the
support void created by widowhood can be hard to fill.
Never-Married Persons
Although the literature has documented that never-married persons are
resourceful in developing support networks throughout their lifetimes (Johnson &
Catalano, 1981), it is unknown whether they receive sufficient support when facing a
major life stressor (Zettel, 2004). Depending on kin such as nieces or nephews is not an
optimal solution for meeting ADL needs. The majority of never-married elders are also
childless; hence, they are likely to lack the two most important sources of instrumental
112

support—spouses and adult children. As noted in the literature, relatives other than
spouses and children make up weaker ties and are not as reliable in providing long-term
instrumental support (Lima & Allen, 2001). However, the relationship between never-
married persons and ADL difficulty has rarely been studied; very few data are available
on the level of ADL support they receive.
In conclusion, divorced, widowed, and never-married persons share a higher risk
of ADL support deficits than married persons due to the lack of a spouse as a primary
care source. Theoretically, the quantity and quality of the support network may be
different among unmarried persons as a result of their different marital histories. For
example, divorce and widowhood may be sudden, but the state of being never married is
long term, which renders never-married persons the opportunity to cultivate support
resources throughout life. However, researchers are often constrained by the small
sample sizes of never-married groups and lump them together into one ―unmarried‖
group with divorced and widowed persons, who, ideally, should make up a separate
group labeled ―formerly married.‖
Adult Children and ADL Support
Adult children are often the second line of defense for older persons combating
difficulties associated with old age (Rossi & Rossi, 1990). Studying the relationship
between care providers and care recipients, Spector, Fleishman, Pezzin, and Spillman
(2000) concluded that the most likely person to provide care for an adult aged 65 and
older is his or her child (41% of all caregivers), specifically a daughter, followed by his
or her spouse (23% of all caregivers). In addition to providing emotional and instrumental
support with daily activities, adult children serve as advocates and interpreters for their
113

parents in dealing with bureaucracies and social services (N. G. Choi, 1994; Shanas &
Sussman, 1977, 1981).
From the literature, researchers know fairly well what adult children do for their
parents, but they do not know how childless older adults make do without children as
care sources. Do childless persons suffer from support deficits? Do they compensate for
their lack of adult children with other support sources? How effective are those
substitutes? The number of adult children as caregivers for the future old (i.e., baby
boomers) will decrease due to the lower fertility rate among baby boomers. The type of
assistance current childless persons receive and from whom may provide a clue for what
lies ahead for the future old.
Gender Differences and ADL Support
Research has consistently shown that women have a larger and more supportive
social network than men and receive more support when in need (S. Allen et al., 1999;
Dykstra, 1990; Wellman, 1992). Davey and colleagues (1999) also found that elderly
women in the United States were twice likely than men to have ADL assistance. They
reported that men, particularly unmarried men, were at elevated risk of having deficits in
care. Unmarried women receive more support from family than unmarried men (Longino
& Lipman, 1981). Frail older mothers are more likely than frail fathers to receive support
as a function of early affection invested on children (Silverstein, Conroy, Wang,
Giarrusso, & Bengtson, 2002). Widows have a more extensive network of support from
family and friends than widowers do (Longino & Lipman, 1981; Mugford & Kendig,
1986; Wister & Strain, 1986), and divorced fathers can lose more supportive resources
than divorced mothers.
114

In sum, from the theoretical and empirical literature, it is clear that resources of
instrumental support are conferred by gender, marital status, and parental status. Those
without spouses may suffer a more fragmented and potentially less stable array of care
options (Freedman, Aykan, Wolf, & Marcotte, 2004), and it is not clear whether adult
children can fill the void left by a lost spouse. Those without spouses and without
children may suffer further reduced resources. Although childless persons have more
active ties with siblings, friends, and other relatives than parents do, no findings have
indicated that active ties guarantee instrumental support (Chappell & Badger, 1989; Lee
& Ihinger-Tallman, 1980). Similarly, unmarried and childless persons tend to rely on
distant kin and friends for companionship and confiding. However, these relationships, a
vital source in meeting the need for sociability, are not necessarily good sources for long-
term ADL assistance (Johnson & Catalano, 1981).
A missing piece in the previous research is whether and how assistive devices
complement or supplement personal care among older adults with disabilities in light of
gender, marital status, and parental status. This knowledge will have implications for
aging policy to meet the care needs of older adults in the near future, as a large proportion
of baby boomers will live to reach the status of oldest old and to experience ADL
difficulty.
Research Questions
Assistive technology is being promoted by policy makers as a long-term care
strategy. However, which sociodemographic factors affect the adaptation and utilization
of assistive devices is not clear. For example, research has not fully addressed the
relationship between having a spouse and/or adult children and the chance of using
115

assistive devices to reduce ADL difficulty. On the one hand, having a close family
member may increase the use of assistive devices through the engagement of formal care
by the family member. On the other hand, the lack of a close family member may lead to
having insufficient personal assistance, which in turn may propel an elder with a
disability to obtain an assistive device to make up for this shortage. In addition to
variations in usage, do assistive devices impede psychological well-being because they
substitute for a personal touch? Or do they enhance psychological well-being because
they enhance the user‘s independence? These questions are not yet answered in the
literature. My study was designed to provide some answers to these questions, as I
studied ADL assistive devices as a component of care provision. Furthermore, I studied
the impact of childlessness on the availability of formal and informal care, and the
entailed outcomes on psychological well-being, among older persons with disability.
I investigated the following research questions:
1. Do parents and childless persons with difficulty walking across a room or
getting into and out of bed differ in their care provision (i.e., their
utilization of assistive devices, formal care, and informal care)?
2. Does care provision affect the psychological well-being of older adults
with difficulty walking across a room or getting into and out of bed?
3. Does childlessness affect the psychological well-being of older adults with
difficulty walking across a room or getting into and out of bed, in the
context of marital status and gender?
116

Method
Data
This study uses data from Health and Retirement Study (HRS) conducted by the
Institute for Social Research at the University of Michigan. HRS is a longitudinal study
with a household panel surveyed every two years. The original sample for the first study
(1992) consists of respondents aged 51 to 61 at time of survey and their spouses
(regardless of spouse‘s age). In 1998, its companion study, the Study of Assets and
Health Dynamics Among the Oldest Old (AHEAD), was integrated into HRS; together
they are the revised Health and Retirement Study. AHEAD has an initial sample of 7,447
respondents aged 70+ (and their spouses, if married, regardless of age). HRS and
AHEAD started to use a uniformed questionnaire after they merged in 1998. The analysis
in this study included four waves of HRS data from 1998 to 2004.
Sample Selection
To examine how ADL difficulty is alleviated by the influences of socio-
demographic factors, I focus on older adults with limitations in ADL. To be included in
this study, a person must have difficulty, prior to using assistive devices or personal
assistance, in performing at least one ADL item included in this study.
Three major predictors in this study are parental status (having ever mothered or
fathered a biological child or not), gender, marital status (married or unmarried; the later
including separated, divorced, widowed, and never married). Childless people compose
only a small portion of the qualified respondents. To achieve a sufficient number of them
for analysis, I pooled participants from four waves of HRS data from 1998 to 2004. Since
HRS contains a panel of subjects who are surveyed repeatedly every two years, a person
117

may be included in this study more than once due to his/her presence in multiple waves if
meeting sampling criteria in those waves. Therefore, the sample unit in this study is not
―a person,‖ but ―an observation‖ of a person. Since the observations are pooled from four
waves of HRS, a person may contribute one to four records to the final analysis.
However, data contained in each observation of the same person should vary to a certain
degree due to advance of the person‘s age, possible changes in marital status, in ADL
condition and self-rated health condition. To account for multiple observations from the
same respondents in the sample design, eventually I applied the statistical procedure of
robust cluster of respondents.
In summary, three criteria were used to select sample, which are:
1. Age —the respondent had to be at least 75 years old at the time of data
collection,
2. Having difficulty walking across a room or getting into or out of bed (the two
ADL items focused on in this study), and
3. Living in a community instead of a nursing home.

Originally 3,021 records from the four waves from 1998 to 2004 met the criteria
for sample selection. However, I excluded 850 cases that had missing values on various
Center for Epidemiologic Studies–Depression (CES-D) items, which were used in HRS
to measure psychological well-being and were outcome variables in this study. According
to HRS documents, in 99.67% of cases of respondents with missing CES-D data, it was
because the respondents were represented by proxy in the interview. Because the CES-D
measures subjective feelings, a proxy cannot answer these questions for the respondent.
118

Excluding participants missing CES-D data from the 3,021 sample pool resulted in 2,171
valid observations. Furthermore, I excluded observations of respondents with missing
data on parental status (n = 48), marital status (n = 8), household assets (n=12), as well as
observations of parents who had outlived all of their children (n = 43). The final sample
set included 2,060 observations.
Comparing the 850 observations excluded from the sample because of missing
CES-D data with those included in the sample, I found that the excluded participants
were more likely to be married (p < .05), to be male, to be older, to have less education,
to be a parent, and to have walking across a room and getting into or out of bed difficulty
(all ps < .001). The implication is that the prevalence of difficulty of walking across and
getting into or out of bed presented in this study may be less severe than in reality among
the population that this sample represents.
On the other hand, comparing the 48 observations with missing data on parental
status with those included in the sample, I found no difference in terms of gender,
education level, marital status, or prevalence of difficulty getting into or out of bed.
However, the group with missing data on parental status was less likely to have difficulty
walking across a room. The implication is that the final sample may lean toward having
slightly more prevalence of difficulty walking across a room than would be found in the
general population with disabilities in this age group. This concern may be partially offset
by the effect from excluding 850 proxy-using observations which renders the sample to
be less prevalent in ADL difficulty in walking across a room.
I excluded observations with missing data on marital status (n = 8), and education
(n = 7), as well as observations of respondents who had outlived all of their children (n =
119

43), instead of imputing their values. This was because the low number of missing values
in these variables did not justify imputation.
Sample Composition and Attrition
This study was based on 2,060 observations from 1,461 respondents. On average,
a single respondent had 1.41 observations in this study. As the range of observations per
respondent was 1 to 4, the mean number of observations fell below its median of 2.5.
This was due to sample attrition and participant death in the HRS. The loss of sample
occurs in every longitudinal study and is particularly common for studies of the older
population. In this study, there was 60% sample loss from wave to wave. A descriptive
analysis showed that those who dropped out in the subsequent waves were not different
from those who stayed in the study in marital status and parental status, but they tend to
be older (83.62 vs. 83.15 in age; p<.01), are more likely to be men (p<.05), have less
education (10.19 vs. 10.53 in years; p<.01), have more difficulty getting in/out of bed
(p<.01), more likely to use assistive devices for getting in or out of bed (p<.05), and
receiving more personal assistance with walking across a room and getting into or out of
bed (ps<.01).
Measures
Definition and Operationalization of Disability
In this study, I defined disability by ADL difficulty. A person who has one or
more ADL difficulty is classified as disabled (World Health Organization, 1993).The
status of difficulty was established through self-report. The HRS questionnaire identified
ADL difficulty with the following question: ―Because of a health or memory problem, do
120

you have any difficulty with X activity?‖ (yes, no, can‘t do, or don‘t do). In this study, I
classified can‘t do and don‘t do as yes (i.e., difficulty in performing that task).
I focused on two items of ADL difficulty: difficulty walking across a room and
difficulty getting in or out of bed. The reason for choosing these two items was twofold:
1. They are the most common ADL difficulties experienced by older adults,
according to the U.S. National Health Interview Survey in 1994 and 1995 (Agree &
Freedman, 2003).
2. They are the two ADL items for which the HRS has data about the use of
equipment or devices to perform the specific ADL task. Assistive device is considered as
a form of care provision in alleviating ADL difficulty (Freedman, Agree, Martin, and
Cornman; 2005).
Definition and Operationalization of Care Provision
The HRS asked whether respondents ―ever use[d] equipment or devices such as
….‖ for the two ADL tasks of walking across a room and getting in or out of bed.
Regardless of whether the respondent used equipment, the questionnaire proceeded to ask
whether the respondent received personal assistance (―Does anyone ever help you with X
activity?‖). There was some overlap in the use of assistive devices and personal
assistance: 23.7% of the observations of walking across a room difficulty used both
assistive devices and personal assistance, and 17.6% of the observations of getting in or
out of bed difficulty used both. Further more, the assistive devices used to aid walking
across a room and getting in and out of bed are often the same. According to HRS data,
the top three assistive devices participants used, in the order of popularity, for walking
across a room are: 1) cane, 2) walker, and 3) wheelchair/cart; for getting in and out of bed
121

are: 1) walker, 2) cane, and 3) Railing. The similarity in the choice of devices justifies a
combined analysis for both ADL condition in the regard of device usage.
I further classified personal care sources to be formal care and informal care. The
formal care or informal care counts the primary care provider only, which was identified
in HRS by the question of ―who helps you most?‖ If the caregiver was a family member,
relative, friend, neighbor, then, it was counted as informal care. If the caregiver was
identified as a paid helper, an employee of an organization or a professional, then, the
care was counted as formal care. It should be noted that in HRS questionnaire, the
identification of care source is not ADL item-specific, which means that it does not
identify which source helps which ADL item. Instead, it is one general question applying
personal care to all ADL item(s) that a particular participant has. Based on this, it is also
justified to analyze personal care source for both walking across a room and getting in
and out of bed together since there is no way to be certain which ADL the care source
assists with.
Definition of Childlessness
In this study, childlessness is defined as the status of never having had a
biological child, regardless whether one has an adopted child or stepchild. This is the
most widely adopted definition of childlessness in aging research (Koropeckyj-Cox,
1998). I chose to limit the definition of childlessness to biological children because of
data constraints in the HRS that do not provide information on adopted children, and also
because of the uncertainty associated with the stepparent–stepchild relationship. Because
the HRS does not survey whether a person has adopted any children, studies using HRS
data have no means of exploring this particular form of parent–child relationship
122

(Koropeckyj-Cox, 1998; Zhang & Hayward, 2001). Information about stepchildren is
available in the HRS but was excluded from this study because of the inherent ambiguity
and diversity so common in the experience of stepparenthood (Cherlin & Furstenberg,
1994). Therefore, the existence or absence of stepchildren is not considered in this study.
It is worth noting that Zhang and Hayward (2001) suggested that stepchildren can be a
potential source of social support, and Koropeckyj-Cox (1998) raised the issue that most
surveys do not make distinctions between biological, adoptive, and stepchildren, which is
indeed a question that deserves greater attention.
In the analysis, permanently childless adults are compared to their peers who are
biological parents with at least one living child. This study does not include those parents
who survived all of their children, due to the assumption that the meaning of parenthood
is different for a person who once had children but lost them all and a person who never
had biological children. The number of parents who survived all of their children is 43
observations (2.1% of the sample). They were excluded from the final sample.
Operationalization of Childlessness
The HRS asks respondents about their parental status when they are first
surveyed. The question is not put to them in the successive waves because the majority of
participants are aged 50 and older (unless that respondent is a much younger spouse of
the primary respondent) and are assumed to have passed the age of childbearing. The
question is worded as follows: ―How many children have you fathered/have you given
birth to? Please don‘t count miscarriages or still-births, or adopted or step-children for
this question.‖
Because the HRS is a panel study, parental status information for most
123

respondents is recorded in the early waves (mostly in the 1995 AHEAD and 1996 HRS
waves; this question was not asked in the 1992 or 1994 HRS waves), with later waves
containing parental status information for new participants only. I combined data from all
waves to form a complete list of the number of biological children of each participant.
The question regarding whether a respondent has ever had a biological child is
followed by one that surveys the number of living children: ―How many of them are still
living?‖ As with the question about having had a biological child, the question about the
number of living children is asked the first time a participant joins the panel but not again
in successive waves. This poses a potential problem of being unable to identify
respondents who have lost all of their children after the first wave of participation in the
HRS. However, the number of parents who outlive their children is small (less than 1%
of the entire HRS sample); therefore, this limitation should not impact my study
significantly.
To compile a complete set of parental status information, I compared the number
of biological children and the number of living children. If the number of biological
children was at least 1, and the number of living children was not 0, then the person did
not meet the criterion of outliving all children, and I recorded the outliving-all-children
variable as 0. If the person did not have a biological child, there was no question about
surviving all children, so the outliving-all-children value was also 0. Only when a person
had biological child(ren) and no living child was he or she counted as having outlived all
children and the value for that variable entered as 1. A person who survived all children
was excluded from the sample for this study, as explained in ―Definition of
Childlessness.‖
124

Definition of Psychological Well-Being
Analogous to the idea that health is not merely the absence of disease,
psychological well-being is not merely the absence of depression. Psychological well-
being is a multidimensional concept that acknowledges the potential coexistence of both
positive and negative feelings. Therefore, when assessing psychological well-being,
researchers should investigate the manifestation of both positive and negative aspects of
emotions.
Almost all previous research on the psychological well-being of childless elders
or widowed older persons has focused on negative aspects (i.e. depression and
loneliness). This study expands the investigative scope to include positive aspects (i.e.
happiness and enjoyment of life). Furthermore, it is a mistake to assume that seemingly
stressful events always lead to stress (Pearlin, Lieberman, Menaghan, & Mullan, 1981),
or that stressors necessarily lead to negative outcomes. Therefore, both positive and
negative outcomes should be measured. It is not hard to imagine that suffering disability
is a stressor, but a person may count his or her blessing when receiving care and feeling
good about being cared for. The manifestation of positive aspects of well-being is
recognized in the form of global happiness, satisfaction with family life, and satisfaction
with friendship in most adopted measures. In the Center for Epidemiologic Studies–
Depression scale (CES-D; Radloff, 1977), they are measured as ―being happy‖ and
―enjoying life.‖
Three indicators of psychological well-being—depression, loneliness, and
sadness—composed the measure of depressive affect in my study. They are traditionally
cited in normative admonitions against childlessness (Koropeckyj-Cox, 2002). Loneliness
125

has been defined as an unpleasant feeling of dissatisfaction with either the number or the
quality of existing social relationships (DeJong-Gierveld, 1987; Koropeckyj-Cox, 1998,
2002; Perlman & Peplau, 1982). Depression is ambiguous in definition and has various
manifestations of feeling low, blue or simply down (Blazer, 1995).
Operationalization of Psychological Well-Being
Psychological well-being is measured in the HRS with the CES-D, a self-report
depression scale for research in the general public, which was designed to measure
depressive symptoms by emphasizing their subjective and affective elements (McDowell
& Newell, 1996; Radloff, 1977). The CES-D has been widely used in research on the
general and clinical populations (Radloff, 1977), and it has good internal consistency
(Hann, Winter, & Jacobsen, 1999). The HRS adopted a 9-item battery of CES-D items,
with each item limited to a dichotomous yes/no response. Six of the nine items indicate
the presence of depression as manifested in negative feelings and somatic symptoms, and
three indicate positive feelings. Positive feelings include being happy, enjoying life, and
having a lot of energy much of the time during the past week; negative feelings include
being depressed, lonely, or sad much of time during the past week; somatic indicators
include feeling that everything is effort, sleeping restlessly, and finding it hard to get
going. The lead-in question for the CES-D in the HRS reads as follows: ―Now think
about the past week and the feelings you have experienced. Please tell me if each of the
following was true for you much of the time during the past week. Much of the time
during the past week …‖ Several published papers have used the CES-D to study mental
health based on HRS data (Gallo, Bradley, Siegel, &Kasl, 2000; Siegel, Bradley, Gallo,
& Kasl, 2003, 2004), and Gallo and colleagues concluded that it has acceptable
126

measurement properties after assessing its reliability and validity using 1992 and 1994
HRS data.
In this study, individual CES-D items were grouped into two indexes: depressive
affect and positive affect of psychological well-being. The following steps were involved
in the transformation: (1) using factor analysis to extract a positive affect factor and a
depressive affect factor from the nine items.; the factor analyses extracted two factors: (a)
depressive affect, which included the three original CES-D items of being depressed,
feeling lonely, and feeling sad; and (b) positive affect, which included the two original
CES-D items of being happy and enjoying life, and (2) adding together items belonging
to the same factor to form an index of depressive affect (0-3 items) and positive affect (0-
2 items).
Although I initially transformed the original discrete dichotomous measure into an
ordinal measure through factor analysis and index-composition process, I eventually
treated depressive affect and positive affect as a dichotomy for the logistic regression
analysis. In the regression, depressive affect was either 0 (no depressive symptoms at all)
or 1 (one to three depressive symptoms). Positive affect was coded as 0 (not being happy
and not enjoying life) or 1 (being happy or enjoying life, or both).
Predictive Variables
Gender was both a categorical variable and a predictor. The predictor marital
status included two categories: married or unmarried. Originally, I divided participants
into three groups as married, formerly married (including separated, divorced, widowed)
and never married. However, due to the small number of never-married in the sample
(n=49 with walking across a room difficulty and n=34 with getting in and out of bed
127

difficulty), I had to group them with the formerly married to be a general group of the
unmarried. I grouped the being separated persons with the divorced, because
conceptually they are similar in situation and neither subgroup had a large sample size.
Control Variables
Five demographic variables were included as control variables. These are: (1) age,
(2) years of education, (3) race (Caucasian =1, others=0), (4) household assets (in
thousand dollars), and (5) self-rated health condition (1= excellent, 2=very good, 3=good,
4=fair, 5=poor).
Statistical Procedure
I first performed a descriptive analysis to gain a picture of demographic
characteristics, the distribution of ADL difficulty and care provision among the sample.
Following that, I applied logistic regression to identify influencing power of predictive
variables on different care provision relatively. Finally, I used logistic regression to
identify the relative effects (odds ratio) of independent variables and care provision on
psychological well-being. To account for multiple observations from the same
respondents in the sample design, I adjusted standard errors by robust cluster of
respondents.
Results
Sample Profile
Participants in this sample were aged 75 to 103 years old, with a mean age of
83.15 and an average educational level of 10.53 years. As expected, in this old-old group
there were more women than men, and the ratio of women to men was almost 3 to 1
128

(71.4% vs. 28.6%, respectively). The respective percentages of parents and childless
persons were 85.7% and 14.3%.
Unmarried persons accounted for 65.1% of the participants. The high proportion
of unmarried persons was due in large part to the number of widowed persons, reflecting
the increase in the likelihood of widowhood in advanced age.
With regard to care provision, 20.4% of respondents did not use assistive devices
or having personal care to alleviate their ADL difficulty. The majority of participants
(54.4%) used some type of assistive device to combat ADL difficulty but did not have
personal care; 2.7% had formal care as their primary personal care source; and 22.5% had
informal care as their primary personal care source, with or without the use of assistive
devices.
Although all participants experienced functional difficulty, 89.4% expressed
being happy and/or enjoying life, and only 58.3% of participants showed depressive
symptoms. The correlation of depressive affect and positive affect was a moderate –.247,
which indicated the coexistence of depressive and positive affects. I summarize this
discussion in Table C1.
129

Table C1. Demographic Characteristics, ADL Difficulty, and ADL Care Provision
Variable Entire Sample
(N = 2,060)
Age (in years) 83.15
Education (in years) 10.53
% Male 28.6
% Caucasian 81.1
% Childless 14.3
% Unmarried 65.1
% With walk difficulty 77.2
% With in/out bed difficulty 49.0
Among entire sample
% No equipment, no personal assistance

20.4
% Have equipment, no personal assistance 54.4
% Formal care as primary care 2.7
% Informal care as primary care 22.5
% With depressive symptoms 58.3
% With positive affect 89.4
Note: ADL = activity of daily living.

Distribution of Care Provision
Table C2 shows that the older a person became, the less likely that he or she was
to go without assistive devices (p < .001) or personal assistance (p < .05). Educational
level and personal care attainment had an inverse relationship—the more education one
had, the less likely that he or she had personal care. With an increase in household assets,
130

the likelihood of having personal assistance also increased. The most noticeable
determinant was marital status. Being unmarried increased the chance of having no
personal care (p < .001). A separate analysis showed that the deficit in personal care for
unmarried persons was mainly attributed to lack of informal care, not formal care (data
not shown).
Table C2. Logistic Regression on Assistive Device Use and Personal Care
Variable Asistive Devices

Personal Care (formal or
informal care)

Model 1 Model 2 Model 3 Model 1 Model 2 Model 3
Age
1.081**** 1.082**** 1.082**** 1.021** 1.022** 1.022**
Gender (male = 1)
0.896 0.856 0.839 0.810 0.725** 0.741**
Education
1.017 1.017 1.017 0.955*** 0.955*** 0.955***
Race (ref = minority)
0.609*** 0.607*** 0.609*** 0.867 0.865 0.862
Household assets
1.000 1.000 1.000 1.002*** 1.002*** 1.002***
Childless (ref = parent)
0.775 0.711* 0.543* 0.799 0.626** 0.820
Single (ref = married)
1.017 1.009 0.954 0.412**** 0.404**** 0.426****
Male childless (Gender
× Childless)
1.351 1.464 2.245** 2.042*
Unmarried childlesss
(Single × Childless)
1.447 0.669
R
2

.0334 .0338 .0344 .0332 .0363
Note: Data are odds ratios.
*p < .10; **p < .05; ***p < .01; ****p < .001 (2-tailed).
Gender and childlessness were not significant in determining care provision at
first glance (Model 1 for both the assistive devices and personal care regressions). But the
introduction of interactive terms for gender and childlessness, and also childlessness and
131

singlehood, revealed differences between men and women in care provision. Childless
women were slightly less likely to use assistive devices (p < .10; Model 2 of the assistive
devices regression), and had lower odds of having personal care than childless men,
whereas childlessness and gender interact to increase the odds of having personal care for
childless men (ps < .05; Model 2 of the personal care regression). Upon the addition of
the interaction of childlessness and singlehood into the regression (Model 3 of both
regressions), marital status did not change the lower usage of devices among childless
women. However, being married increased the odds of having personal care for childless
women, as the odds ratio of childlessness went from lower odds to no difference, as
shown in Models 2 and 3 of the personal care regression. It is also worth noting that,
except for childless men, who had higher odds of having personal care, fathers actually
had lower odds of having personal care when compared with women (p < .05; Model 2 of
the personal care regression).
Psychological Well-Being
Table C3 shows that for older adults with disabilities, a determinant factor for
depressive affect was educational level—having more education reduced the odds of
having depressive symptoms (p < .001). Being unmarried or rating one‘s own health as
poor were detrimental to psychological well-being—they significantly increased the odds
of having depressive affect (ps < .001). Being unmarried increased the odds of having
depressive affect, but it did not hold discernable power over positive affect. Self-rated
health conditions had a strong relation with positive affect (p < .001), such that the better
one stated his or her health was, the more likely he or she expressed feeling happy and
enjoying life. Using assistive devices alleviated problems with ADLs, which entailed a
132

lower odds of having depressive affect. With regard to personal care, when compared to
informal care as the primary care source, having no personal care or having formal care
as the primary care source did not have a differential effect on depressive affect or
positive affect. The interactive terms of (a) gender and childlessness and (b) childlessness
and singlehood were not significant, and they did not increase the explanation power of
the models. Hence, they were left out of the final regression models.
Table C3. Logistic Regression of Psychological Well-Being on Demographic
Characteristics of Older Adults With Difficulty
Variable Psychological Well-Being
Depressive Affect Positive Affect
Model 1 Model 2 Model 1 Model 2
Age 0.994 0.999 0.984 0.981
Gender (male = 1) 1.016 1.013 1.050 1.047
Education 0.911**** 0.913**** 1.015 1.012
Race (ref = minority) 1.068 1.040 0.574** 0.587**
Household assets 0.999 0.999 0.999 0.999
Childless (ref = parent) 1.027 1.015 1.435 1.451
Single (ref = married) 1.448*** 1.497*** 1.009 1.012
Health (1 = excellent, 5 = poor) 1.335**** 1.350**** 0.610**** 0.602****
Using assistive devices 0.669*** 1.342
Personal care (ref = informal
care as primary source)
No personal care

0.837

1.017
Formal care
0.958 0.583
R
2
.0489 .0535 .0373 .0404
Note: Data are odds ratios.
**p < .05; ***p < .01; ****p < .001.
133

Discussion and Limitations
In this study I investigated the use of assistive devices among older adults with at
least one of two ADL difficulties—walking across a room or getting into or out of bed—
in the context of parental status, marital status, and gender. I also studied whether
different types of care provision—usage of assistive devices, formal care as primary care,
and informal care as primary care—impact psychological well-being differently, and
whether childlessness affects the psychological well-being of older adults with
disabilities.
I found that the use of personal assistance was mainly contingent upon marital
status, with unmarried persons being more likely to lack informal care as a primary care
source. Neither gender nor childlessness by itself was a significant factor in determining
care provision, but gender and childlessness interacted to slightly reduce the odds of
using assistive devices for childless women and to increase the use of personal care for
childless men. Having a lower odds of using assistive devices may indirectly increase
childless women‘s odds of having depressive affect, as using assistive devices can lead to
a lower odds of having depressive affect. The slightly more personal care that childless
men enjoy does not necessarily extend to benefits in psychological well-being, as
personal care, regardless of its availability or its source, does not directly affect
psychological well-being in either depressive affect or positive affect.
It is somewhat surprising and seems counterintuitive that childless men, married
or not, have slightly higher odds of having personal care. One may assume that they
suffer from double jeopardy (lack adult children, and having a smaller social network
than women) with regard to personal care resources, or even triple jeopardy, if they do
134

not have a spouse. One possible explanation is that never-married persons account for
41.0% of childless unmarried men. Never-married childless men may get their support
sources in place early on as bachelors, especially if relatives believe they need to be cared
for. However, this still does not explain why childless men have more personal care than
childless women. A possibility is that, as researchers have suspected, the social support of
women does not necessarily translate to ADL support.
In this study, I was able to test the interaction between being unmarried and being
childless. The interaction was not significant, perhaps due to the small size (and thus low
statistical power) of the childless group. Nevertheless, the odds ratios did point to a
higher reliance on assistive devices (odds ratio = 1.447) and a lower chance of having
personal care (odds ratio = 0.669), as presented in Table C2.
I did not find a relationship between personal care and psychological well-being,
regardless of whether a participant had no personal care at all, formal care as a primary
help source, or informal care as a primary help source. This seems to disagree with the
evidence that social and instrumental support reduce stress. I argue that the potential
negative consequences of having no personal care were alleviated by the high usage of
assistive devices in this sample. In fact, the prevalence of assistive device usage was
higher than that of personal care among respondents (54.4% vs. 22.5%, respectively).
The no-difference finding between formal care and informal care on
psychological well-being may be due to the fact that the number of participants who
received formal care in this sample (only 2.7%) was too small to be statistically powerful
even if some difference did exist. It may also be explained by the possibility that using
formal care helps older adults with disabilities to overcome their ADL difficulty and
135

relieves them from feelings of being a burden to family and friends. Besides, many other
factors (e.g., self-perceived health, coping skills, perceived control, motivations, etc.)
determine the relationship between disability and perceived quality of life, and its
reflection on psychological well-being (Bowling, Seeta, Morris, & Ebrahim, 2007). Care
provision is just one of these factors.
It is when one is disabled that he or she likely needs family the most to go about
daily life. Although childless women are slightly less likely to use assistive devices,
childless persons with disabilities generally do not receive less care or have worse
psychological well-being. This may indicate a payoff for childless persons‘ long-term
involvement with the children of siblings, neighbors, and close friends. Many childless
older adults have strong and durable family ties with parents, siblings, cousins, nieces,
and nephews. Those relationships may be sources of care.
These findings conclude that the types of care childless persons receive do not
affect their psychological well-being. However, considering the outcomes of care
provision beyond psychological well-being (e.g., toward the desire to age in place),
researchers must ask whether type of care matters in the end. Their lower likelihood of
using assistive devices plus their lower odds of having personal care may explain why
unmarried childless women are more likely than their other elderly counterparts to live in
institutions (Boaz & Muller, 1994; Freedman, 1996; Koropeckyj-Cox, 2007) and have
higher mortality rates (Hurt, Ronsmans, & Thomas, 2006). Apparently, other outcomes
besides psychological well-being are important measures when considering the effects of
parental status, marital status, and gender among older adults with disabilities.
136

The large number of aging baby boomers signifies a potentially huge demand for
long-term care in the coming decades, particularly in the domain of ADL assistance.
Meanwhile, baby boomers have a smaller pool of informal care resources than did their
parents. Some have questioned whether family members can continue to be the primary
support sources for boomers and provide them with ADL assistance when they are in
need. Judging by demographic trends, baby boomers are more likely than previous
generations to be unmarried and childless. It is projected that 25% of baby boomers will
not have a spouse or a child when entering old age (Wachter, 1997). This translates to a
significant proportion of future elders who will lack two of the most primary sources for
daily task assistance—spouses and adult children. Therefore, it is compelling for
researchers to better understand how the existence of a spouse and/or a child affects
assistance with ADL difficulty and well-being in various domains. But as to what extent
the deficit in personal care can be made up by assistive devices demands further research.
The findings in this study point out the promising prospect of promoting assistive devices
and technology as a long-term care tool. The understanding of the usage of personal care
and assistive devices will arm policy makers with proper knowledge to develop long-term
care policies and help professionals predict the need for formal care down the road.

137

Chapter Five: General Conclusion and Discussion
Parenthood anchors the so-called ―normal expectable life‖ for most people
(Dykstra & Hagestad, 2007b, p. 1276). Deviation from the path of parenthood often
invites public pity (if one is involuntarily childless) or punishment (if voluntarily
childless). The pressure of conforming was particularly harsh for currently old childless
persons during their childbearing years. With the various social changes that have
occurred in the past four decades, in particular the acceptance of various forms of family,
public censure for childlessness has relaxed, resulting in greater acceptance of
childlessness and childless people. However, despite research findings that repeatedly
refute the notion of childless persons as more depressed and isolated than parents,
negative perceptions of childlessness persist, fueled by higher institutionalization rates
among unmarried older women.
Two shortcomings common to existing research make results on childlessness and
old-age well-being inconclusive and uncertain: (a) treating childlessness as a static status
and not considering possible changes in its meaning and impact felt in significant life
events, and (b) focusing on negative aspect of childlessness at the exclusion of possible
positive rewards. This dissertation added contextual variables—commonly experienced
stressful events in various life stages from midlife to the oldest old age—and also
measured both negative and positive affects of psychological well-being to enrich
researchers‘ understanding of childlessness and psychological well-being from middle
age to the very old years. In view of the increasing childless population among baby
boomers and the continuing trend among younger generations to regard parenthood as a
choice and not a mandate, it is imperative that scholars revisit the issue of childlessness
138

and old-age well-being in order to update their understanding of the currently old
childless population and plan for future childless elders.
Based on findings from the three studies included in this dissertation, I conclude
that childlessness by itself does not pose a negative threat to psychological well-being.
This is in agreement with most previous research findings. However, one natural outcome
of childlessness is the lack of biological grandchildren. The second study of this
dissertation (Paper B, on new retirees) showed that grandchildren have a positive
influence on grandparents‘, and in particular grandfathers‘, transitions to retirement. The
lack of a biological grandchild clearly indicates the loss of a potential source of happiness
and enjoyment of life that most grandparents enjoy.
It should be emphasized that the lack of a negative effect of childlessness on
psychological well-being does not mean that children are not beneficial to parents.
Rather, the findings point to the fact that taking the ―deviant‖ path of childlessness,
whether voluntarily or not, does not necessarily render childless persons at a greater
disadvantage than parents in old age. This can easily be explained through the lens of
human beings‘ adaptive ability. Everyone adjusts to nonoptimal situations in life the best
they can. Childlessness is a long-term status; childless people adapt to it, regardless of
whether it is by choice of by circumstances. Meanwhile, they may appreciate some
advantages over parents, such as greater flexibility with finances or time. Childless
persons arrange their alternative resources to compensate for any possible void stemming
from the lack of children in their lives. Furthermore, the prospect of having no adult
children as old-age ―insurance‖ may prompt them to start planning for their old age
earlier; resources saved from not having to rear children can be used for this purpose.
139

This is supported by research findings that childless persons (and in particular childless
women) are more active at taking charge of their situations and choosing their options for
care and housing (Rubin-Terrado, 1994). This active planning may give childless persons
a sense of autonomy, which, in turn, may manifest as positive affect when these people
are faced with disability in old age. As Bowling and colleagues (2007) noted, perceived
control contributes positively to psychological well-being.
Having no children does not affect depressive affect across the life course.
However, not having a close relationship with at least one of their children is detrimental
to parents compared with parents with a close relationship with at least one child. This
does not affect depressive affect, but it has a strong impact on positive affect. When
combining this fact with the finding that grandchildren bring joy and happiness to
grandparents, one can infer that offspring bring about happiness and enjoyment in later
life. In contrast, spouses are more effective at helping elders fend off loneliness, sadness,
or depression, as evidenced by my findings that being married reduces depressive affect
for both men and women from midlife to the oldest old years in the face of stressful
events such as caregiving, retirement, and disability.
The difference that a close parent–child relationship makes on parents‘ positive
affect supports the argument that it is important to differentiate intragroup variation
among parents. Some researchers have proposed differentiating the origin of
childlessness as voluntary or involuntary. Due to the constraints of the HRS, which has
no data on the origin of childlessness, I was unable to test whether one‘s self-claimed
origin of childlessness is important to psychological well-being in old age. But I argue
that the origin of childlessness does not divide childless persons as much as quality of the
140

parent–child relationship does for parents. The main reason is that the pain from an
unsatisfactory relationship with one‘s child is present and acute, whereas the pain of
involuntary childlessness may be dulled by time and alleviated through adaptation, as
evidenced by the blurred boundaries of voluntary or involuntary childlessness among
childless participants in previous studies.
The importance of considering parental status within marital status has been
emphasized by other scholars as well (e.g., Zhang & Hayward , 2001). Due to the small
size of the childless group and some parent subgroups (e.g., parents with no close
children) in the present studies, particularly after they have been further divided by
gender, it was not statistically feasible to test the interaction term of childlessness and
marital status in the study of childlessness and caregiving or the study of childlessness
and retirement transition. But in all three studies, the effects of childlessness were teased
out from the impact of marital status.
My findings also support the need to consider the intersection of gender, parental
status, and marital status. Gender does make a difference for older adults‘ psychological
well-being. It was obvious in the new retirees study that being married helped male
retirees transition to retirement with less depressive affect and more positive affect,
whereas it only affected female retirees‘ depressive affect, and to a lesser extent for
women than for men. In addition, grandchildren are influential to grandfathers‘ positive
affect, but I did not see the same effect for grandmothers. Also, among oldest old adults
with disability, childless women had slightly lower odds of using assistive devices,
whereas childless men had a slightly greater chance of having personal care than
childless women. These findings indicate that childlessness has different implications for
141

men and women. They point to the importance of considering gender differences in
assessing the outcomes of being childless. Due to the small number of childless
respondents in this sample, I was unable to test a three-way interaction among gender,
parental status, and marital status. This remains a venue worth pursuing.
The chance of parents outliving all their children increases with the parents‘ age.
Fewer than 1% of respondents in the entire sample had outlived all their children, but this
number was 2.1% among the adults aged 75 and older in the study on older adults with
disability. In view of the uniqueness of their situation, and the fact that they became
childless only after having been parents, parents who had survived all of their children
occupy a special category in the study of parental status among the oldest old.
The size of the childless population is expected to continue to grow across age
groups. Among the baby boom generation, this growth is due to higher infertility rates
than among previous generations. Among the younger generation, it is due to the use of
highly effective contraceptives and the greater societal acceptance of singlehood and
childlessness. Each decade an increasing proportion of young adults remains childless by
choice, whereas delayed marriage and fertility control result in more couples remaining
inadvertently childless.
The findings from this dissertation have implications for public policy. The
retirement transition study, which found that grandchildren impact grandparents‘ (and
especially grandfathers‘) positive affect, supports intergenerational relationship
programs. The care provision study, which found that childlessness does not have a direct
negative impact on the psychological well-being of older adults with disabilities but may
have an indirect impact through less use of assistive devices among childless women, has
142

an implication for social services programs targeting childless women. The finding that
use of assistive devices reduces depressive affect is an affirmation of current public
policies promoting assistive devices as a long-term care strategy (one such policy is the
New Freedom Initiative, signed into law in 2001 by President George W. Bush). The
finding also calls for researchers to pay greater attention to the potential inequality of
device adaptation among groups of different socioeconomic statuses.
As Dysktra and Hagestad (2007b) pointed out, the topic of childlessness is
conspicuously absent in studies and major publications on families in later life. In the
past 30 years, the efforts spent by researchers on understanding childless older adults
have not been commensurate with the size of the childless population. This dissertation,
by focusing on older adults without children, contributes to researchers‘ knowledge about
families and social networks and sheds new light on gender contrasts in later life. It
provides insights into dynamics of the life course, social integration, and inequalities in
psychological well-being according to parental status, marital status, and gender.

143

BIBLIOGRAPHY
AARP. (2001, July). In the middle: A report on multicultural boomers coping with family
and aging issues. Washington, DC: Author.

Acitelli, L. K., & Antonucci, T. C. (1994). Gender differences in the link between marital
support and satisfaction in older couples. Journal of Personality and Social
Psychology, 67, 688–698.

Agree, E. M., & Freedman, V. A. (2003). A comparison of assistive technology and
personal care in alleviating disability and unmet need. The Gerontologist, 43,
335–344.

Alexander, B. B., Rubinstein, R. L., Goodman, M., & Luborsky, M. (1992). A path not
taken: A cultural analysis of regrets and childlessness in lives of older women.
The Gerontologist, 32, 618–626.

Allen, K. R., Blieszner, R., & Roberto, K. A. (2000). Families in the middle and later
years: A review and critique of research in the 1990s. Journal of Marriage and
Family, 62, 911–926.

Allen, S., Goldscheider, F., & Ciambrone, D. (1999). Gender roles, marital intimacy, and
nomination of spouse as primary caregiver. The Gerontologist, 39, 150–158.

Andrews, F. M., Abbey, A., & Halman, L. J. (1991). Stress from infertility, marriage
factors, and subjective well-being of wives and husbands. Journal of Health and
Social Behavior, 32, 238–253.

Anson, O., Antonovsky, A., Sagy, S., & Adler, I. (1989). Family, gender, and attitudes
toward retirement. Sex Roles, 20, 355–368.

Antonucci, T. C., & Akiyama, H. (1987). An examination of sex differences in social
support among older men and women. Sex Roles, 17, 737–749.

Atchley, R. (1974). The meaning of retirement. Journal of Communications, 24, 97-101.

Atchley, R. (1975). Adjustment to loss of job and retirement. International Journal of
Aging and Human Development, 6, 17–27.

Atchley, R. (1976). Selected social and psychological differences between men and
women in later life. Journal of Gerontology, 31, 204–221.

Austrom, M. G., Perkins, A. J., Damush, T. M., & Hendrie, H. C. (2003). Predictors of
life satisfaction in retired physicians and spouses. Social Psychiatry and
Psychiatric Epidemiology, 38, 134–141.
144

Barker, J. C. (2002). Neighbors, friends, and other nonkin caregivers of community-
living dependent elders. Journal of Gerontology: Social Sciences, 57B, S158–
S167.

Barnes, H., & Parry, J. (2004). Renegotiating identity and relationships: Men and
women‘s adjustments to retirement. Ageing & Society, 24, 213–233.

Beckett, A. (2006, May 3). Who is brooding now? The Guardian. Retrieved October 23,
2007, from
http://www.guardian.co.uk/science/2006/may/03/genderissues.familyandrelations
hips

Beckman, L. J., & Houser, B. B. (1982). The consequences of childlessness on the social-
psychological well-being of older women. Journal of Gerontology, 37, 243–250.

Belgrave, L. L., & Haug, M. R. (1995). Retirement transition and adaptation: Are health
and finances losing their effects? Journal of Clinical Geropsychology, 1, 43–66.

Bender, K. A., & Jivan, N. A. (2005, February). What makes retirees happy? (Issue Brief
No. 28). Boston: Boston College, Center for Retirement Research.

Bengtson, V. L. (2001). Beyond the nuclear family: The increasing importance of
multigenerational bonds. Journal of Marriage and Family, 63, 1–16.

Blake, J. (1979). Is zero preferred? American attitudes toward childlessness in the 1970s.
Journal of Marriage and Family, 41, 245–257.

Blazer, D. G. (1995). Depression. In J. E. Birren (Ed.), The encyclopedia of aging (2nd
ed., pp. 265–266). New York: Springer.

Boaz, R. F., & Hu, J. (1997). Determining the amount of help used by disabled elderly
persons at home: The role of coping reources. Journal of Gerontology: Social
Sciences, 52B, S317–S324.

Boaz, R. F., & Muller, C. F. (1994). Predicting the risk of ―permanent‖ nursing home
residence: The role of community help as indicated by family helpers and prior
living arrangements. Health Service Research, 29, 391–414.

Bowling, A. (1994). Social networks and social support among older people and
implications for emotional well-being and psychiatric morbidity. International
Review of Psychiatry, 9, 447–459.

Bowling, A. (2005). Ageing well. Maidenhead, England: Open University Press.

145

Bowling, A., Seeta, S., Morris, R., & Ebrahim, S. (2007). Quality of life among older
people with poor functioning: The influence of perceived control over life. Age
and Ageing, 36, 310–315.

Brody, E. M., Hoffman, C., Kleban, M. H., & Shoonover, C. B. (1989). Caregiving
daughters and their local siblings: Perceptions, strains, and interactions. The
Gerontologist, 29, 529–538.

Broverman, I., Vogel, S., Broverman, D., Clarkson, F., & Rosenkrantz, P. (1972). Sex
role stereotypes: A current appraisal. Journal of Social Issues, 28, 59–78.

Brubaker, T. H. (1990). Families in later life: A burgeoning research area. Journal of
Marriage and Family, 52, 960–981.

Bulcroft, R., & Teachman, J. (2003). Ambiguous constructions: Development of a
childless or childfree life course. In M. Coleman & L. H. Ganong (Eds.),
Handbook of contemporary families: Considering the past, contemplating the
future (pp. 116–135). London: Sage.

Burdick, D. C. (2007). Gerontechnology. In J. E. Birren (Ed.), Encyclopedia of
gerontology (Vol. 1, pp. 619-630). St. Louis, MO: Elsevier.

Burr, W. R., Leigh, G. K., Day, R. D., & Constantine, J. (1979). Symbolic interaction and
the family. In W. R. Burr, R. Hill, F. I. Nye, & I. L. Reiss (Eds.), Contemporary
theories about the family (Vol. 11, pp. 42–111). New York: Free Press.

Cain, M. (2001). The childless revolution. Cambridge, MA: Perseus.

Calasanti, T. M. (1996). Gender and life satisfaction in retirement: An assessment of the
male model. Journal of Gerontology: Social Sciences, 51B, S18–S29.

Callan, V. J. (1987). The personal and marital adjustment of mothers and of voluntarily
and involuntarily childless wives. Journal of Marriage and Family, 49, 847–856.

Campbell, A., Converse, P. E., & Rodgers, W. L. (1976). The quality of American life.
New York: Russell Sage Foundation.

Campbell, E. (1985). The childless marriage. London: Tavistock.

Cantor, M. H. (1979). Neighbors and friends: An overlooked resource in the informal
support system. Research on Aging, 1, 434–463.

Carstensen, L. L. (1995). Evidence for a life-span theory of socioemotional selectivity.
Current Directions in Psychological Sciences, 4, 151–156.

146

Chapman, N. J. (1989). Gender, marital status, and childlessness of older persons and the
availability of informal assistance. In M. D. Peterson & D. L. White (Eds.),
Depression and the social environment (pp. 395–405). Montreal, Quebec,
Canada: McGill-Queen‘s University Press.

Chappell, N. L., & Badger, M. (1989). Social isolation and well-being. Journal of
Gerontology: Social Sciences, 44, S169–S176.

Chappell, N. L., & Blandford, A. (1991). Informal and formal care: Exploring the
complementarity. Ageing and Society, 11, 299–315.

Cherlin, A. J., & Furstenberg, F. F., Jr. (1994). Stepfamilies in the United States: A
reconsideration. Annual Review of Sociology, 20, 359–381.

Choi, H., & Marks, N. F. (2006). Transition to caregiving, marital disagreement, and
psychological well-being. Journal of Family Issues, 27, 1701–1722.

Choi, N. G. (1994). Patterns and determinants of social service utilization: Comparison
of the childless elderly and elderly parents living with or apart from their children.
The Gerontologist, 34, 353–362.

Choi, N. G. (1996). The never-married and divorced elderly: Comparison of economic
and health status, social support, and living arrangement. Journal of
Gerontological Social Work, 26(1/2), 3–25.

Choi, N. G. (2002). Self-defined retirement status and engagement in paid work among
older working-age women: Comparison between childless women and mothers.
Sociological Inquiry, 72(1), 43–71.

Cicirelli, V. G. (1983). A comparison of helping behavior to elderly parents of adult
children with intact and disrupted marriage. The Gerontologist, 23, 616–681.

Connidis, I. A., & McMullin, J. A. (1992). Getting out of the house: The effect of
childlessness on social participation and companionship in later life. Canadian
Journal on Aging, 11, 370–386.

Connidis, I. A., & McMullin, J. A. (1993). To have or have not: Parent status and the
subjective well-being of older men and women. The Gerontologist, 33, 630–636.

Connidis, I. A., & McMullin, J. A. (1994). Social support in older age: Assessing the
impact of marital and parent status. Canadian Journal on Aging, 13, 510–527.

Connidis, I. A., & McMullin, J. A. (1996). Reasons for and perceptions of childlessness
among older persons. Journal of Aging Studies, 10, 205–222.

147

Covey, H. (1981). A reconceptualization of continuity theory: Some preliminary
thoughts. The Gerontologist, 21, 628–633.

Davey, A., Femia, E. E., Shea, D. G., Zarit, S. H., Sundstrom, G., Berg, S., et al. (1999).
How many elders receive assistance? A cross-national comparison. Journal of
Aging and Health, 11, 199–220.

Dehle, C., & Weiss, R. L. (1998). Sex differences in prospective associations between
marital quality and depressed mood. Journal of Marriage and Family, 60, 1002–
1011.

DeJong-Gierveld, J. (1987). Developing and testing a model of loneliness. Journal of
Personality and Social Psychology, 53, 119–128.

Dooghe, G. (1994). Social and health aspects. In G. J. Stolinitz (Ed.), Social aspects and
country reviews of population aging: Europe and North America (pp. 9–36). New
York: United Nations.

Dorfman, L. T. (1989). Retirement preparation and retirement satisfaction in the rural
elderly. Journal of Applied Gerontology, 8, 432–450.

Dorfman, L. T. (1995). Health conditions and perceived quality of life in retirement.
Health and Social Work, 20, 192–200.

Dorfman, L. T. (2002). Retirement and family relationships: An opportunity in later life.
Generations, 26(2), 74–79.

Dykstra, P. A. (1990). Disentangling direct and indirect gender effects on the supportive
network. In C. P. Knipscheer & T. C. Antonucci (Eds.), Social network research:
Substantive issues and methodological questions (pp. 55–65). Amsterdam, The
Netherlands: Swets & Zeitlinger.

Dykstra, P. A. (2006). Off the beaten track: Childlessness and social integration in late
life. Research on Aging, 28, 749–767.

Dykstra, P. A., & Hagestad, G. O. (2007a). Childlessness and parenthood in two
centuries: Different roads, different maps? Journal of Family Issues, 28, 1518–
1532.

Dykstra, P. A., & Hagestad, G. O. (2007b). Roads less taken. Journal of Family Issues,
28, 1275–1310.

Easterlin, R. A. (2003, March). Building a better theory of well-being (Discussion Paper
No. 742). Bonn, Germany: Institute for the Study of Labor (IZA)

148

Ekerdt, D. J. Bossé, R., & LoCastro, J. S. (1983). Claims that retirement improves health.
Journal of Gerontology, 38, 231–236.

Elder, G. H., Jr., & Johnson, M. K. (2003). The life course and aging: Challenges,
lessons, and new directions. In R. A. Settersten, Jr. (Ed.), Invitation to the life
course: Toward new understandings of later life (pp. 49–81). New York: Wiley.

Erikson, E. H. (1968). Generativity and ego integrity. In B. Neugarten (Ed.), Middle age
and aging (pp. 85-87). Chicago: University of Chicago Press.

Fisher, B. (1991). Affirming social value: Women without children. In R. Maines (Ed.),
Social organization and social process (pp. 87–104). New York: Aldine de
Gruyter.

Forste, R. (2002). Where are all the men? A conceptual analysis of the role of men in
family formation. Journal of Family Issues, 23, 579–600.

Fouquereau, E., Fernandez, A., & Mullet, E. (2001). Evaluation of determinants of
retirement satisfaction among working and retired people. Social Behavior and
Personality, 29, 777–786.

Freedman, V. (1996). Family structure and the risk of nursing home admission. Journal
of Gerontology: Social Sciences, 51B, S61–S69.

Freedman, V. A., Agree, E. M., Martin, L. G., & Cornman, J. C. (2005). Trends in the
use of assistive technology and personal care for late-life disability, 1992–2001.
The Gerontologist, 46, 124–127.

Freedman, V. A., Aykan, H., Wolf, D. A., & Marcotte, J. E. (2004). Disability and home
care dynamics among older unmarried Americans. Journal of Gerontology: Social
Sciences, 59B, S25–S33.

Furstenberg, F. F. (2005). Banking on families: How families generate and distribute
social capital. Journal of Marriage and Family, 67, 809–821.

Gallo, W. T., Bradley, E. H., Siegel, M., & Kasl, S. V. (2000). Health effects of
involuntary job loss among older workers: Findings from the Health and
Retirement Survey. Journal of Gerontology: Social Sciences, 55B, S131–S140.

Gatz, M., Bengtson, V. L., & Blum, M. J. (1990). Caregiving families. In J. E. Birren &
K. W. Schaie (Eds.), Handbook of the psychology of aging (3rd ed., pp. 404–426).
San Diego, CA: Academic Press.

149

George, L. K., Fillenbaum, G. G., & Palmore, E. B. (1984). Sex differences in the
antecedents and consequences of retirement. Journal of Gerontology, 39, 364–
371.

George, L. K., & Maddox, G. (1977). Subjective adaptation to loss of the work role: A
longitudinal study. Journal of Gerontology, 39, 456–462.

Gerson, K. (1985). Hard choices: How women decide about work, career and
motherhood. Berkeley: University of California Press.

Gerstel, N., & Gallagher, S. K. (1993). Kinkeeping and distress: Gender, recipients of
care and work-family conflict. Journal of Marriage and Family, 55, 598–607.

Giarrusso, R., Silverstein, M., & Bengtson, V. (1996). Family complexity and the
grandparent role. Generations, 20(1), 17–23.

Gibson, D., & Mugford, S. (1986). Expressive relations and social support. In H. L.
Kendig (Ed.), Ageing and families: A social networks perspective (pp. 63-84).
Boston: Allen & Unwin.

Gilligan, C. (1982). In a different voice. Cambridge, MA: Harvard University Press.

Gironda, M., Lubben, J. E., & Atchison, K. A. (1999). Social networks of elders without
children. Journal of Gerontological Social Work, 31, 63–84.

Glenn, N. D., & McLanahan, S. (1981). The effects of offspring on the psychological
well-being of older adults. Journal of Marriage and Family, 43, 409–421.

Glenn, N. D., & Weaver, C. N. (1979). A note on family situation and global happiness.
Social Forces, 57, 960–967.

Goldscheider, F. K. (1990). The aging of the gender revolution: What do we know what
do we need to know? Research on Aging, 12, 531–545.

Goode, W. J. (1960). A theory of role strain. American Sociological Review, 25, 483–
496.

Gove, W. R. (1984). Gender differences in mental and physical illness: The effects of
fixed roles and nurturant roles. Social Science & Medicine, 19, 77–84.

Greene, M. E., & Biddlecom, A. E. (2000). Absent and problematic men: Demographic
accounts of male reproductive roles. Population and Development Review, 26,
81–115.

150

Gustman, A. L., Mitchell, O. S., & Steinmeier, T. L. (1995). Retirement measures in the
Health and Retirement Study. Journal of Human Resources, 30, S57–S83.

Gutmann, D. L. (1975). Parenthood, key to the comparative study of the life cycle. In N.
Datan & L. Ginsberg (Eds.), Life-span developmental psychology: Normative life
crises (pp. 167–184). New York: Academic Press.

Hagestad, G. O. (2007). Pathways to childlessness: A life course perspective. Journal of
Family Issues, 28, 1487–1517.

Hann, D., Winter, K., & Jacobsen, P. (1999). Measurement of depressive symptoms in
cancer patients: Evaluation of the Center for Epidemiological Studies Depression
Scale (CES-D). Journal of Psychosomatic Research, 46, 437–443.

Hardy, M. A., & Quadagno, J. (1995). Satisfaction with early retirement: Making choices
in the auto industry. Journal of Gerontology: Social Sciences, 50B, S217–S228.

Havighurst, R. J. (1972). Developmental tasks and education. New York: McKay.

Himes, C. L. (1992). Future caregivers: Projected family structures of older persons.
Journal of Gerontology: Social Sciences, 47, S17–S26.

Houser, B. B., Berkman, S. L., & Beckman, L. J. (1984). The relative rewards and costs
of childlessness for older women. Psychology of Women Quarterly, 8, 395–398.

Hurt, L. S., Ronsmans, C., & Thomas, S. L. (2006). The effect of number of births on
women‘s mortality: Systematic review of the evidence for women who have
completed their child-bearing. Population Studies, 60, 55–71.

Isaksson, K. & Johansson, G. (2000). Adaptation to continued work and early retirement
following downsizing: Long-term effects and gender differences. Journal of
Occupational and Organizational Psychology, 73(2), 241-256.
Jeffries, S., & Konnert, C. (2002). Regret and psychological well-being among
voluntarily and involuntarily childless women and mothers. International Journal
of Aging and Human Development, 54, 89–106.

Jerrome, D., & Wenger, G. C. (1999). Stability and change in late-life friendships.
Ageing & Society, 19, 661–676.

Johnson, C. L., & Catalano, D. J. (1981). Childless elderly and their family support. The
Gerontologist, 21, 610–618.

Johnson, C. L., & Troll, L. (1992). Family functioning in late life. Journal of
Gerontology: Social Sciences, 47, S66–S72.
151

Jordan, J. V., Kaplan, A. G., Miller, J. B., Stiver, I. P., & Surrey, J. L. (1991). Women ’s
growth in connection. New York: Guilford Press.

Kemkes-Grottenhaler, A. (2003). Postponing or rejecting parenthood? Results of a survey
among female academic professionals. Journal of Biosocial Science, 35, 213–226.

Kendig, H., Dykstra, P. A., van Gaalen, R. I., & Melkas, T. (2007). Health of aging
parents and childless individuals. Journal of Family Issues, 28, 1457–1486.

Kim, J. E., & Moen, P. (2001). Is retirement good or bad for subjective well-being?
Current Directions in Psychological Science, 10, 83–86.

Kim, J. E., & Moen, P. (2002). Retirement transitions, gender, and psychological well-
being: A life-course, ecological model. Journal of Gerontology: Psychological
Sciences, 57B, P212–P222.

Kim, S., & Feldman, D. C. (2000). Working in retirement: The antecedents of bridge
employment and its consequences for the quality of life in retirement. Academy of
Management Journal, 43, 1195–1210.

Knight, B. G., Robinson, G. S., Flynn Longmire, C.V., Chun, M., Nakao, K., & Kim, J.
H. (2002). Cross cultural issues in caregiving for persons with dementia: Do
familism values reduce burden and distress? Ageing International, 27(3), 70–94.

Koropeckyj-Cox, T. (1998). Loneliness and depression in middle and old age: Are the
childless more vulnerable? Journal of Gerontology: Social Sciences, 53B, S303–
S312.

Koropeckyj-Cox, T. (2002). Beyond parental status: Psychological well-being in middle
and old age. Journal of Marriage and Family, 64, 957–971.

Koropeckyj-Cox, T. (2007). Characteristics of older childless persons and parents.
Journal of Family Issues, 28, 1362–1414.

Kreider, R. M. (2003). Adopted children and stepchildren: 2000: Census 2000 special
reports (CENSR-6RV). Washington, DC: U.S. Census Bureau.
Laditka, J. N., & Laditka, S. B. (2001). Adult children helping older parents. Research
on Aging, 23, 429–456.

Lakey, B., & Cohen, S. (2000). Social support theory and measurement. In S. Cohen, L.
G. Underwood, & B. Gottlieb (Eds.), Social support measurement and
interventions: A guide for health and social scientists (pp. 29-52). Oxford: Oxford
University Press.
152

Larsson, K., & Silverstein, M. (2004). The effects of marital and parental status on
informal support and service utilization: A study of older Swedes living alone.
Journal of Aging Studies, 18, 231–244.

Lawton, M. P., Moss, M., Kleban, M. H., Glicksman, A., & Rovine, M. (1991). A two-
factor model of caregiving appraisal and psychological well-being. Journal of
Gerontology: Social Sciences, 46, S181–S189.

Lee, G. R. (1978). Marriage and morale in later life. Journal of Marriage and Family, 40,
131–139.

Lee, G. R., & Ihinger-Tallman, M. (1980). Sibling interaction and morale: The effects of
family relations on older people. Research on Aging, 2, 367–391.

Lima, J. C., & Allen, S. M. (2001). Targeting risk for unmet need: Not enough help
versus no help at all. Journal of Gerontology: Social Sciences, 56B, S302–S310.

Litwak, E. (1985). Helping the elderly. New York: Guilford Press.

Longino, C. F., Jr., & Lipman, A. (1981). Married and spouseless men and women in
planned retirement community: Support network differentials. Journal of
Marriage and Family, 43, 169–177.

Loomis, L. S., & Booth, A. (1995). Multigenerational caregiving and well-being. Journal
of Family Issues, 16, 131–148.

Marks, N. F. (1995). Midlife marital status differences in social support relationships
with adult children and psychological well-being. Journal of Family Issues, 16, 5–
28.

Marks, S. R. (1977). Multiple roles and role strain. American Sociological Review, 42,
921–936.

May, E. T. (1995). Barren in the promised land: Childless Americans and the pursuit of
happiness. New York: Basic Books.

McDowell, I., & Newell, C. (1996). Measuring health: A guide to rating scales and
questionnaires (2nd ed.). New York: Oxford University Press.

McGoldrick, A. M., & Cooper, C. L. (1994). Health and aging as factors in the retirement
experience. European Work and Organizational Psychologist, 4, 1–20.

McMullin, J. A., & Marshall, V. W. (1996). Family, friends, stress and well-being: Does
childlessness make a difference? Canadian Journal on Aging, 15, 355–373.
153

MedicineNet.com. (2003). Webster’s new world medical dictionary (6th ed.). London:
Macmillan Reference Books.

Messeri, P., Silverstein, M., & Litwak, E. (1993). Choosing optimal support groups: A
review and reformulation. Journal of Health and Social Behavior, 34, 122–137.

Miall, C. E. (1986). The stigma of involuntary childlessness. Social Problems, 33, 268–
282.

Midanik, L. T., Soghikian, K., Ransom, L. J., & Tekawa, I. S. (1995). The effect of
retirement on mental health and health behaviors: The Kaiser Permanente
Retirement Study. Journal of Gerontology: Social Sciences, 50B, S59–S61.

Miller, B. (1990). Gender differences in spouse caregiver strains: Socialization and role
expectations. Journal of Marriage and Family, 52, 311–321.

Mitrani, V. B., Lewis, J. E., Feaster, D. J., Czaja, S. J., Eisdorfer, C., Schulz, R., et al.
(2006). The role of family functioning in the stress process of dementia
caregivers: A structural family framework. The Gerontologist, 46, 97–105.

Moen, P. (2001). The gendered life course. In R. L. Binstock & L. K. George (Eds.),
Handbook of aging and the social sciences (5th ed., pp. 179-196). San Diego, CA:
Academic Press.

Mugford, S., & Kendig, H. L. (1986). Social relations: Networks and ties. In H. L.
Kendig (Ed.), Ageing and families: A social networks perspective (pp. 38-59).
Boston: Allen & Unwin.

Murray, J. (1973, October). Retirement history study report, No. 4. Social Security
Bulletin, 3–23.

Mutran, E. J., Reitzes, D. C., & Fernandez, M. E. (1997). Factors that influence attitudes
toward retirement. Research on Aging, 19, 251–273.

National Academy on an Aging Society. (2000, June). Who are young retirees and older
workers? Washington, DC: Author.

National Center for Health Statistics. (1990). Advance report of final natality statistics,
1998. Monthly Vital Statistics Report, 39(4).

National Institute on Aging. (2007). Menopause. Retrieved May 8, 2007, from
http://www.niapublications.org/agepages/menopause.asp

154

Neugarten, B. L. (1968). The awareness of middle age. In B. L. Neugarten (Ed.), Middle
age and aging: A reader in social psychology (pp. 93–98). Chicago: University of
Chicago Press.

Nieva, V. F. (1985). Work and family linkages. Women and Work: An Annual Review, 1,
162–190.

Norris, F. H., & Kaniasty, K. (1996). Received and perceived social support in times of
stress: A test of the social support deterioration deterrence model. Journal of
Personality and Social Psychology, 71, 498–511.

Older Women‘s League. (1989). Failing America ’s caregivers: A status report on women
who care. Washington, DC: Author.

Palmore, E., Fillenbaum, G., & George, L. (1984). Consequences of retirement. Journal
of Gerontology, 39, 109–116.

Patsios, D., & Davey, A. (2005). Formal and informal community care for older adults.
In M. L. Johnson (Ed.), The Cambridge handbook of age and ageing (pp. 597-
604). Cambridge, England: University Press.

Pearlin, L. I., & Aneshensel, C. S. (1994). Caregiving: The unexpected career. Social
Justice Research, 7, 373–390.

Pearlin, L. I., Lieberman, M. A., Menaghan, E. G., & Mullan, J. T. (1981). The stress
process. Journal of Health and Social Behavior, 22, 337–356.

Pearlin, L. I., Pioli, M. F., & McLaughlin, A. E. (2001). Caregiving by adult children:
Involvement, role disruption, and health. In R. H. Binstock & L. K. George
(Eds.), Handbook of aging and social sciences (5th ed., pp. 238–255). San Diego,
CA: Academic Press.

Perlman, D., & Peplau, L. A. (1982). Theoretical approaches to loneliness. In L. A.
Peplan & D. Perlman (Eds.), Loneliness: A sourcebook of current theory,
research, and therapy (pp. 123–134). New York: Wiley Interscience.

Pinquart, M., & Schindler, I. (2007). Changes of life satisfaction in the transition to
retirement: A latent-class approach. Psychology and Aging, 22, 442–455.

Pinquart, M., & Sörensen, S. (2001). Gender differences in self-concept and
psychological well-being in old age: A meta-analysis. Journal of Gerontology:
Psychological Sciences and Social Sciences, 56, P195-P213.

Pinquart, M., & Sörensen, S. (2006). Gender differences in caregiver stressor, social
resources, and health: An updated meta-analysis. Journal of Gerontology:
155

Psychological Sciences, 61B, P33–P45.

Preston, S. H. (1992). Cohort succession and the future of the oldest old. In R. M.
Suzman, D. P. Willis, & K. G. Manton (Eds.), The oldest old (pp. 50-57). New
York: Oxford University Press.

Price, K. F., & Joo, F. (2005). Exploring the relationship between marital status and
women‘s retirement satisfaction. International Journal of Aging and Human
Development, 61, 37–55.

Pruchno, R. A., & Resch, N. L. (1989). Husbands and wives as caregivers: Antecedents
of depression and burden. The Gerontologist, 29, 159–165.

Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for research in the
general population. Applied Psychological Measurement, 1, 385–401.

Reitzes, D. C., & Mutran, E. J. (2002). Self-concept as the organization of roles:
Importance, centrality, and balance. Sociological Quarterly, 43, 647–667.

Reitzes, D. C., & Mutran, E. J. (2004). The transition to retirement: Stages and factors
that influences retirement adjustment. International Journal of Aging and Human
Development, 59, 63-84.

Reitzes, D. C., Mutran, E. J., & Fernandez, M. E. (1996). Does retirement hurt well-
being? Factors influencing self-esteem and depression among retirees and
workers. The Gerontologist, 36, 649–656.

Richardson, V. (1999). Women and retirement. Journal of Women and Aging, 11(2/3),
49–66.

Richardson, V., & Kilty, K. M. (1991). Adjustment to retirement: Continuity vs.
discontinuity. International Journal of Aging and Human Development, 33, 151-
169.

Richardson, V., & Kilty, K. M. (1995). Gender differences in mental health before and
after retirement: A longitudinal analysis. Journal of Women and Aging, 7(1/2),
19–36.

Ross, C. E. (1995). Reconceptualizing marital status as a continuum of social attachment.
Journal of Marriage and Family, 57, 129–140.

Ross, C. E., & Huber, J. (1985). Hardship and depression. Journal of Health and Social
Behavior, 26, 312–327.

156

Ross, C. E., & Mirowsky, J. (2002). Family relationship, social support and subjective
life expectancy. Journal of Health and Social Behavior, 43, 469–489.

Ross, C. E., Mirowsky, J., & Goldsteen, K. (1990). The impact of the family on health:
The decade in review. Journal of Marriage and Family, 52, 1059–1078.

Rossi, A., & Rossi, P. (1990). Of human bonding: Parent –child relations across the life
course. New York: Aldine de Gruyter.

Rubinstein, R. L. (1987). Childless elderly: Theoretical perspectives and practical
concerns. Journal of Cross-Cultural Gerontology, 2, 1–14.

Rubin-Terrado, M. (1994). Social supports and life satisfaction of older childless women
and mothers living in a nursing home. Unpublished doctoral dissertation,
Northwestern University, Evanston, IL.

Russo, N. (1976). The motherhood mandate. Journal of Social Issues, 32, 143–179.

Schlossberg, N. K. (1995). Counseling adults in transitions: Linking practice with theory.
New York: Springer.

Seccombe, K., & Lee, G. R. (1986). Gender differences in retirement satisfaction and its
antecedents. Research on Aging, 8, 426–440.

Seeman, T. E., & Crimmins, E. (2001). Social environment effects on health and aging:
Integrating epidemiologic and demographic approaches and perspectives. Annals
of the New York Academy of Sciences, 954, 88–117.

Shanas, E. (1979). Social myths as hypothesis: The case of family relations of old people.
The Gerontologist, 19, 3–9.

Shanas, E., & Sussman, M. B. (Eds.). (1977). Family, bureaucracy and the elderly.
Durham, NC: Duke University Press.

Shanas, E., & Sussman, M. B. (1981). The family in later life: Social structure and social
policy. In R. W. Fogel, E. Hatfield, S. B. Kiesler, & E. Shanas (Eds.), Aging:
Stability and change in the family (pp. 211–231). New York: Academic Press.

Sieber, S. D. (1974). Toward a theory of role accumulation. American Sociological
Review, 39, 567–578.

Siegal, M., Bradley, E. H., Gallo, W. T., & Kasl, S. V. (2003). Impact of husbands‘
involuntary job loss on wives‘ mental health, among older adults. Journal of
Gerontology: Social Sciences, 58B, S30–S37.

157

Siegal, M., Bradley, E. H., Gallo, W. T., & Kasl, S. V. (2004). The effect of spousal
mental and physical health on husbands‘ and wives‘ depressive symptoms, among
older adults. Journal of Aging and Health, 16, 398–423.

Silverstein, M., & Bengtson, V. L. (1991). Do close parent-child relations reduce the
mortality risk of older parents? Journal of Health and Social Behavior, 32, 382–
395.

Silverstein, M., Conroy, S. J., Wang, H., Giarrusso, R., & Bengtson, V. (2002).
Reciprocity in parent–child relations over the adult life course. Journal of
Gerontology: Social Sciences, 57B, S3–S13.

Social Security Administration. (2005). Social Security fact sheets. Retrieved November
22, 2005, from http://www.ssa.gov/pressoffice/factsheets/basicfact-alt.pdf

Spector, W. D., Fleishman, J. A., Pezzin, L. E., & Spillman, B. C. (2000). The
characteristics of long-term care users (AHRQ Publication No. 00-0049).
Rockville, MD: Agency for Healthcare Research and Policy.

Spillman, B. C., & Pezzin, L. E. (2000). Potential and active family caregivers: Changing
networks and the ‗sandwich generation.‘ Milbank Quarterly, 78, 347–374.

Spitze, G., Logan, J. R., Joseph, G., & Lee, E. (1994). Middle generation roles and the
well-being of men and women. Journal of Gerontology: Social Sciences, 49,
S107–S116.

Stoller, E P. (1990). Men as helpers: The role of sons, relatives, and friends. The
Gerontologist, 30, 228–235.

Stoller, E. P., & Pugliesi, K. L. (1989). Other roles of caregivers: Competing
responsibilities or supportive resources. Journal of Gerontology: Social Sciences,
44, S231–S238.

Stone, R., Cafferata, G. L., & Sangl, J. (1987). Caregivers of the frail elderly: A national
profile. The Gerontologist, 27, 616–626.

Strawbridge, W. J., & Wallhagen, M. I. (1991). Impact of family conflict on adult child
caregivers. The Gerontologist, 31, 770–777.

Stuckey, J. C., Neundorfer, M. M., & Smyth, K. A. (1996). Burden or well-being: The
same coin or related currency? The Gerontologist, 36, 686–693.

Stull, D., Bowman, K., & Smerglia, V. (1994). Women in the middle: A myth in the
making? Family Relations, 43, 319–324.

158

Szinovacz, M. E. (1998). Grandparents today: A demographic profile. The Gerontologist,
38, 37–52.

Szinovacz, M. E., & Davey, A. (2001). Retirement effects on parent-adult child contacts.
The Gerontologist, 41, 191–200.

Szinovacz, M. E., & Davey, A. (2005). Retirement and marital decision making: Effects
on retirement satisfaction. Journal of Marriage and Family, 67, 387–398.

Szinovacz, M. E., & Davey, A. (2006). Effects of retirement and grandchild care on
depressive symptoms. International Journal of Aging and Human Development,
62, 1–20.

Thornton, A. (1989). Changing attitudes toward family issues in the United States.
Journal of Marriage and Family, 51, 873–893.

Thornton, A., & Young-DeMarco, S. (2001). Four decades of trends in attitudes toward
family issues in the United States: The 1960s through the 1990s. Journal of
Marriage and Family, 63, 1009–1037.

Thornton, A., & Freedman, D. (1982). Changing attitudes toward marriage and single
life. Family Planning Perspectives, 14, 297–303.

Toseland, R. W., Smith, G. C., & McCallion, P. (1995). Supporting the family in elder
care. In G. C. Smith, S. S. Tobin, E. A. Roberstson-Tchabo, & P. W. Power
(Eds.), Strengthening aging families: Diversity in practice and policy (pp. 3–24).
Thousand Oaks, CA: Sage.

Umberson, D., Williams, K., & Sharp, S. (2000). Medical sociology and health
psychology. In C. E. Bird, P. Conrad, & A. M. Fremont (Eds.), Handbook of
medical sociology (5th ed., pp. 353-364). Upper Saddle River, NJ: Prentice Hall.

Veevers, J. E. (1973). The social meanings of parenthood. Psychiatry, 36, 291–310.

Veevers, J. E. (1980). Childless by choice. Toronto, Ontario, Canada: Butterworths.

Voydanoff, P., & Donnelly, B. W. (1999). Multiple roles and psychological distress: The
intersection of the paid worker, spouse, and parental roles with the role of the
adult child. Journal of Marriage and Family, 61, 725–738.

Wachter, K. W. (1997). Kinship resources for the elderly. Philosophical Transition:
Biological Sciences, 352, 1811–1817.

Walker, A. J., Pratt, C. C., & Eddy, L. (1995). Informal caregiving to aging family
members: A critical review. Family Relations, 44, 402–411.
159

Wallsten, S. S. (2000). Effects of caregiving, gender, and race on the health, mutuality,
and social supports of older couples. Journal of Aging and Health, 12, 90–111.

Warr, P., Butcher, V., Robertson, I., & Callinan, M. (2004). Older people‘s well-being as
a function of employment, retirement, environmental characteristics, and role
preferences. British Journal of Psychology, 95, 297–324.

Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief
measures of positive and negative affect: The PANAS scales. Journal of
Personality and Social Psychology, 54, 1063–1070.

Wellman, B. (1992). Which types of ties and networks provide what kinds of social
support. Advances in Group Processes, 9, 207–235.

Wenger, G. C. (1997). Successful ageing from the point of view of old people: Findings
from the Bangor Longitudinal Study of Ageing. Baseline, 64, 16–20.

Wenger, G. C., Dykstra, P.A., Melkas, T. & Knipscheer, K. C. P. M. (2007). Social
embeddedness and late-life support networks. Journal of Family Issues, 28, 1419-
1456.

Wenger, G. C., Scott, A., & Patterson, N. (2000). How important is parenthood?
Childlessness and support in old age in England. Ageing & Society, 20, 161–182.

Williamson, R. C., Rinehart, A. D., & Blank, T. O. (1992). Early retirement: Promises
and pitfalls. New York: Plenum Press.

Wister, A. V., & Strain, L. (1986). Social support and well-being: A comparison of older
widows and widowers. Canadian Journal on Aging, 5, 205–220.

World Health Organization. (1993). International classification of impairments,
disabilities, and handicaps. Geneva, Switzerland: Author. (Original work
published 1980)

Wu, A. S., Tang, C. S., & Yan, E. C. (2005). Post-retirement voluntary work and
psychological functioning among older Chinese in Hong Kong. Journal of Cross-
Cultural Gerontology, 20, 27–45.

Yael, K. (1984/1985). Predictors of retirement satisfaction: A path model. International
Journal of Aging and Human Development, 20, 113–121.

160

Zarit, S. H., Reever, K. E., & Bach-Peterson, J. (1980). Relatives of the impaired elderly:
Correlates of feelings of burden. The Gerontologist, 20, 649–655.

Zettel, L. A. (2004). Aging alone: Do the social support resources of never-married
individuals place them at risk? Unpublished doctoral dissertation, University of
California, Irvine.

Zhang, Z., & Hayward, M. (2001). Childlessness and the psychological well-being of
older persons. Journal of Gerontology: Social Sciences, 56B, S311–S320.

161

Appendix A: Measures of Psychological Well-Being in HRS

Now think about the past week and the feelings you have experienced. Please
tell me if each of the following was true for you much of the time during the
past week.
Much of the time during the past week, you felt depressed.
(Would you say yes or no?)

FELT ACTIVITIES WERE EFFORTS
(Now think about the past week and the feelings you have experienced. Please
tell me if each of the following was true for you much of the time during the
past week.)
(Much of the time during the past week...)
You felt that everything you did was an effort.
(Would you say yes or no?)

WAS SLEEP RESTLESS W/IN PREV WK
(Now think about the past week and the feelings you have experienced. Please
tell me if each of the following was true for you much of the time during the
past week.)
(Much of the time during the past week...)
Your sleep was restless.
(Would you say yes or no?)

WAS R HAPPY W/IN PREV WK

162

(Now think about the past week and the feelings you have experienced. Please
tell me if each of the following was true for you much of the time during the
past week.)
(Much of the time during the past week...)
You were happy.
(Would you say yes or no?)

LONELINESS FELT W/IN PREV WK
(Now think about the past week and the feelings you have experienced. Please
tell me if each of the following was true for you much of the time during the
past week.)
(Much of the time during the past week...)
You felt lonely.
(Would you say yes or no?)

ENJOYED LIFE W/IN PREV WK
(Now think about the past week and the feelings you have experienced. Please
tell me if each of the following was true for you much of the time during the
past week.)
(Much of the time during the past week...)
You enjoyed life.
(Would you say yes or no?)

FELT SAD W/IN PREV WK
(Now think about the past week and the feelings you have experienced. Please
163

tell me if each of the following was true for you much of the time during the
past week.)

(Much of the time during the past week...)
You felt sad.
(Would you say yes or no?)

FELT UNMOTIVATED W/IN PREV WK
(Now think about the past week and the feelings you have experienced. Please
tell me if each of the following was true for you much of the time during the
past week.)
(Much of the time during the past week...)
You could not get going.
(Would you say yes or no?)

FELT FULL OF ENERGY W/IN PREV WK
(Now think about the past week and the feelings you have experienced. Please
tell me if each of the following was true for you much of the time during the
past week.)
(Much of the time during the past week...)
You had a lot of energy.
(Would you say yes or no?)

Abstract (if available)

Abstract The childless population in the United States is growing fast, accounting for approximately one fifth of population aged 65 and older. The combination of longevity, marital status changes, and childlessness has led to the projection that 30% of the US Caucasian population aged 70 to 85 in 2030 will not have a living spouse or a living biological child. Surprisingly, there is relatively little documentation about the lives of childless elders. Childless people have been conspicuously ignored in social sciences, even in very pertinent fields such as adult development, aging, the life course, and the family.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Longitudinal change in active life expectancy: the longitudinal studies of aging 1984-2000

PDF

The measurement, life course patterns, and outcomes of intergenerational ambivalence among parent-adult child dyads

PDF

Life course implications of adverse childhood experiences: impacts on elder mistreatment, subjective cognitive decline, and caregivers' health

PDF

Children 's migration and the financial, social, and psychological well-being of older adults in rural China

PDF

Intergenerational transmission of values and behaviors over the family life course

PDF

Internet communication use, psychological functioning and social connectedness at older ages

PDF

Self-perceptions of Aging in the Context of Neighborhood and Their Interplay in Late-life Cognitive Health

PDF

Intergenerational support between grandparents and grandchildren in rural China and its effect on the psychological well-being of older adults

PDF

What leads to a happy life? Subjective well-being in Alaska, China, and Australia

PDF

Forensic markers of physical elder abuse in medical and community contexts: implications for criminal justice interventions

Asset Metadata

Creator Chang, Echo Win-Hu (author)

Core Title Childlessness and psychological well-being across life course as manifested in significant life events

School Leonard Davis School of Gerontology

Degree Doctor of Philosophy

Degree Program Gerontology

Publication Date 03/03/2008

Defense Date 02/06/2008

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag caregiving,childlessness,Disability,family and aging,OAI-PMH Harvest,Retirement

Language English

Advisor Silverstein, Merril (committee chair), Biblarz, Timothy J. (committee member), Wilber, Kathleen H. (committee member)

Creator Email echochan@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m1037

Unique identifier UC1290728

Identifier etd-Chang-20080303 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-43685 (legacy record id),usctheses-m1037 (legacy record id)

Legacy Identifier etd-Chang-20080303.pdf

Dmrecord 43685

Document Type Dissertation

Rights Chang, Echo Win-Hu

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu