Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Normative obligations and parental care in social context
(USC Thesis Other)
Normative obligations and parental care in social context
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
NORMATIVE OBLIGATIONS AND PARENTAL CARE IN SOCIAL CONTEXT
by
Daphna Gans
________________________________________________________
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 2007
Copyright 2007 Daphna Gans
ii
Dedication
To my husband, George Gans, for his true love and continuous and
endless support in every possible way. Without you, this would not be possible.
To my children, who inspire me and fill my life with joy, and who
graciously share my energy and time with this greedy project. To Jonathan, for
teaching me to tenaciously finish what I started and for modeling to me how
exciting learning can be; And to Ella, for teaching me to appreciate and enjoy
every day in life and for showing me that learning can take many forms.
To my parents, Ruti and Nir Serlin, for always believing in me and for
being the best role models a child can hope for.
To my grandparents, Leah and Yizchak Zohar, Luba and Moshe Serlin,
and my husband’s grandmother, Andree Clement, for inspiring me to study
aging.
iii
Acknowledgements
I would like to thank the members of my dissertation committee for their
thoughtful and valuable guidance and continuous support: Dr. Vern Bengtson,
Dr. Timothy Biblarz, Dr. Eileen Crimmins, and Dr. Merril Silverstein. I am
grateful to each and every one of my committee members for believing in me
and for their mentorship and generosity. I am honored to have had the
opportunity to learn from such remarkable scholars. Special thanks to Dr. Merril
Silverstein, my advisor, for his valued mentorship and for allowing me to learn
by exposure to and observation of his distinguished research and writing
abilities. I would also like to extend special thanks to my previous mentor from
Michigan State University, Dr. Lawrence Schiamberg for his continued support
and interest in my professional development; and to my first mentor and dear
friend, Dr. Ariela Lowenstein from the University of Haifa, for exposing me to
gerontology and for providing me with a remarkable professional and personal
role model. Finally, I would like to recognize the support of the National
Institute on Aging for this research from grants #R01-AG07977 and #T32-
AG00037.
iv
Table of Contents
Dedication ...............................................................................................................ii
Acknowledgements ................................................................................................ iii
List of Tables ...........................................................................................................v
List of Figures........................................................................................................vii
Abstract ..................................................................................................................ix
Preface ...................................................................................................................xi
Chapter One: Introduction, Background, and Conceptual Framework .................. 1
Chapter Two: Data Description: The Longitudinal Study of Generations
(LSOG) ...................................................................................................... 39
Chapter Three: Norms of Filial Responsibility Across Time and Generations ...... 45
Chapter Four: How Parents Influence Their Children to Provide Support to
Them in Old Age........................................................................................ 91
Chapter Five: Congruence and Discrepancy between Filial Norms and Filial
Norms and Filial Behavior: The Role of Religiosity ................................. 138
Chapter Six: Discussion and Conclusions.......................................................... 176
Bibliography........................................................................................................ 187
v
List of Tables
Table 3.1: Age and filial norms at four times of measurement
(Observations=4,527) ......................................................................58
Table 3.2: Distribution of explanatory variables................................................62
Table 3.3: Maximum likelihood estimates of three-level growth curve
modeling redicting filial norms across 15 years within family
context (N=1,627)............................................................................69
Table 3.4: Variance decomposition from three-level analysis of filial norms.....78
Table 4.1: Descriptive statistics: adult children (N=379) and Mothers
(N=237) .........................................................................................106
Table 4.2: Unstandardized, standardized, and significance levels for
measurement model (Standard errors in parentheses);
(N=379) .........................................................................................117
Table 4.3: Unstandardized, standardized, and significance levels for
structural equation model for the mediating variables
(Standard errors in parentheses); (N=379)....................................121
Table 4.4: Unstandardized, standardized, and significance levels for
structural equation model for the dependent variables
(Standard errors in parentheses); (N=379)....................................124
Table 5.1: Descriptive statistics (N=475) ........................................................152
Table 5.2: Description of sample by religious affiliation..................................152
Table 5.3: Latent class models of intergenerational relationships using
four dichotomous indicators of filial behavior and norms ...............161
Table 5.4: Latent class probabilities and latent class distributions for
constrained three-class model.......................................................162
vi
List of Tables (continued)
Table 5.5: Summary of multinomial logistic regression analysis for
variables predicting class-type affiliation for dissonant and
uninvolved in reference to committed supporters ..........................168
vii
List of Figures
Figure 1.1: Conceptual framework: Ecological dyadic intergenerational
model for understanding factors affecting parental support...........10
Figure 3.1: Growth curves for filial norms by age predicted for male and
female (N=1,627)...........................................................................73
Figure 3.2: Growth curves for filial norms by age predicted for earlier-born
generation (g2) vs. later-born generation (g3) (N=1,627) ...............74
Figure 3.3: Growth curves for filial norms by age by education level
(N=1,627) .......................................................................................76
Figure 3.4: Growth curves for filial norms by age by caregiving experience
(N=1,627) .......................................................................................77
Figure 3.5: Growth curves for filial norms by age predicted by three models
(N=1,627) .......................................................................................81
Figure 3.6: Growth curve for a hypothetical average male who is going
through generational succession at age 60 ....................................83
Figure 3.7: Growth curves of two hypothetical average male subjects;
The first, a male from generation 2 (earlier born) who started
the 90s at age 62. The second, a male from generation 3
(later-born) who started the 90s at age 40......................................86
Figure 4.1: Conceptual and empirical causal model.......................................101
Figure 4.2: Measurement model: Standardized coefficients...........................118
Figure 4.3: Measurement model: Standardized coefficients...........................126
Figure 5.1: Hypothetical classes of congruence and discrepancy based on
the classification of individuals on the two dimensions of filial
norms filial behavior......................................................................149
Figure 5.2: BIC statistics for models with 1 through 8 classes........................162
Figure 5.3: Classification error statistic for models with 1 through 8 classes..163
viii
List of Figures (continued)
Figure 5.4: Latent class conditional probabilities for class 1: Dissonant.........164
Figure 5.5: Latent class conditional probabilities for class 2: Uninvolved .......165
Figure 5.6: Latent class conditional probabilities for class 3: Committed
Supporters....................................................................................165
Figure 5.7: Distribution of the three class-types by religious affiliation ...........166
Figure 5.8: Log odds ratio for membership in the dissonant or uninvolved
class-types versus the committed supporters class-type..............169
ix
Abstract
This dissertation aimed to explain how the various contexts of family
and generation membership, religious affiliation and religiosity, and historical
period, shape the degree to which adult children feel obligated to, and actually
support, their aging parents. More specifically, the objectives were: (1) to
examine change across the lifespan in the strength with which individuals feel
adult children should be committed to parental care; (2) to examine how parents
influence the amount of support they will receive in old age from their adult
children; (3) to examine patterns of congruence or discrepancy between filial
obligation to parental care and actual filial behavior of parental care; and (4) to
investigate the role of religiosity in predicting such patterns. Analyses were
performed using data from the USC Longitudinal Study of Generations
(LSOG)—a multi-wave multi-level four generation study. Multi-level growth
curve analysis from the first study (N = 4,527 time-varying observations over
four time points of 1,627 individuals from 333 families) showed that filial norms
are malleable and change over the adult lifespan. Filial norms peak at midlife
and weaken in old age, especially after the death of both parents, when
generational succession takes place. Structural equation models from the
second study (N = 379 mother-child dyads) demonstrated that parents might
shape their children’s supportive behavior through various mechanisms
including strong emotional bonds, demonstration of the desired caring behavior
earlier in life, and exchange mechanisms. Latent class analyses in the third
x
study (N = 475) demonstrated that a discrepant pattern of dissonant individuals-
-reporting strong obligations yet low actual levels of support--was most
prevalent (52%). The remaining subjects divide equally into two opposing types
of congruent patterns: the uninvolved--those reporting weak norms and low
levels of care, and the committed supporters--those reporting strong
commitment and high levels of support. Multinomial regressions analyses
showed that non-religious adult children were almost three times more likely to
be non committed to, and uninvolved in, parental care. Taken together, this
dissertation demonstrated that intergenerational processes such as
commitment to parental care and actual care should be studied within micro-
and macro-level contexts over time.
xi
Preface
The role of adult children in providing long-term support to their aging
parents and the normative obligations that guide parental support have aroused
much interest in social gerontology and family studies. However, the effects of
the multiple micro- and macro- contexts within which such intergenerational
norms develop and are enacted into actual exchange of support are not well
understood. Current and future demographic trends indicate that the need for
prolonged periods of parental support is growing and will continue to grow. A
continued decline in mortality at the end of the life span, combined with the
aging of the “baby boom” cohort, a significantly relatively large cohort, will result
in the aging of the population. Because at older ages chronic disabling
diseases become more prevalent, the need for family care will continue to grow.
At the same time, various changes in the demography of family life may
threaten the availability of adult children to fulfill these growing needs for help
and support. Declining fertility rates suggests that the elderly will have fewer
children to draw on for support. In addition, unprecedented rates of divorce and
women’s labor force participation may contribute to a weakening of kin
availability.
In such climate, it is imperative to understand what motivates adult
children to support their aging parents, how the normative obligations that guide
parental care develop, and what conditions contribute to—or impede—the
enactment of such norms into actual supportive behavior. This dissertation
xii
aims at contributing to a better understanding of the complex and possibly
reciprocal relationships between the familial, cultural, and religious contexts and
the degree to which adult children feel obligated and actually provide support to
their aging parents over time. More specifically, the objectives of this
dissertation are: (1) to examine the strength with which individuals feel adult
children should be committed to care for aging parents, and how this value
changes across the lifespan, over several decades of historical time, between
successive generations, and in relation to one’s generational position in one’s
family; (2) to examine whether parents can influence the amount of support they
will be receiving in old age from their adult children by affecting their children’s
motivations, normative obligations, or actual supportive behaviors throughout
the life course; (3) to examine the patterns of congruence or discrepancy
between filial norms--what people believe one should do to support one’s
parents, and filial behavior—what people actually do to support their aging
parents; and (4) to investigate the role of religion and religiosity in predicting
such patterns of discrepancy or congruence between norms and actions.
The dissertation addresses these topics using the USC Longitudinal
Study of Generations (LSOG)—a multi-wave multi-level four generations study.
The dissertation is guided by an ecological multi-level dyadic model. As any
other study, this dissertation cannot address all the determinants and the
interrelationships implied by a multi-level contextual model. However, it
addresses several of these complex relationships and raises awareness to the
xiii
importance of considering the complex contexts within which parental support
takes place.
The Format of this Dissertation
This dissertation has six chapters. The first chapter includes an
introduction, a discussion of the importance of the topic, a conceptual
framework, and a literature review. The second chapter presents a discussion
of the data set used for all of the analyses in this dissertation. The following
three chapters present three empirical papers, each presented with its specific
background and literature review as well as a method section. The sixth
chapter presents a summary of the findings, implications for future studies, and
conclusions, and ties the three research papers together as well as to the
conceptual framework that guided this dissertation.
1
Chapter One: Introduction, Background, and Conceptual Framework
A. Introduction
The role of adult children in the provision of long-term care, support and
health-crisis decisions to their aging parents has recently become an area of
multidisciplinary interest. In most cultures, families have traditionally been
responsible for caring for their older members. In the past, however, extended
care occurred infrequently due to low longevity (Zarit & Reid, 1994). Current
and future trends, however, indicate that the need for prolonged periods of
family caregiving is growing and will continue to grow. A sharp decline in
mortality rates over the past century and a continued decline in mortality at the
end of the life span, combined with the aging of the “baby boom” cohort, a
significantly relatively large cohort, will result in the aging of the population as
the number of older persons in the population as well as in the proportion of
older adults in the total population continue to grow (Siegel, 1993). Although
the majority of older adults can live independently or with minimal assistance, a
substantial number of older adults have disabilities and therefore require care.
Because at older age chronic disabling diseases become more prevalent, and
because the rates of severe disability increase with age, the need for family
care and other forms of long-term care is growing (Zarit & Reid, 1994).
The belief that contemporary families do not provide care for their aging
members and the “myth of alienation” (Shanas, 1979), according to which older
adults are alienated from their families, have not been supported by years of
2
research (Brody & Brody, 1989; Cicirelli, 1988, Troll, 1986). In fact, families
always have been and continue to be the main provider of long-term health and
social support to their aging members (Brody & Brody, 1989). Furthermore,
families are viewed by policymakers as social units that are ready, willing, and
able to meet the growing needs of the elderly population, and to provide help to
aging family members at times of need (Parrott & Bengtson, 1999). Most of
family caregiving is performed by spouses, and mainly wives (Stone, Cafferta,
Lee & Sangl, 1987). When a spouse is absent or unable to provide care, adult
children, and typically adult daughters, are next “in line of defense” (Zarit &
Reid, 1994, p. 240). Future trends indicate that family caregiving will become
an increasingly demanding responsibility for adult children (Cicirelli, 1990). It is
estimated that at least 25% of adult daughters may be expected to take on
parental care at varying intensity at some point in their lives (Sorenson & Zarit,
1996). In fact, the caregiving role has become a normative developmental task
in the life course of women (Silverstein & Parrott, 2001; Sorenson & Zarit,
1996), as well as a family developmental task (Zarit & Reid, 1994).
B. Importance of this Study
As parental caregiving becomes a normative part of the life course of
middle aged persons, and particularly women (Silverstein & Parrott, 2001;
Sorenson & Zarit, 1996; Zarit & Reid, 1994), and as the needs of the older
population are growing (Siegel, 1993), special attention should be given to the
ability of adult children to take on the filial role. Previous research has
3
documented the willingness of families in general and adult children in
particular, to care for aging family members (Brody & Brody, 1989; Cicirelli,
1988, 1981; Stone et al., 1987; Troll, 1986). However, various changes in the
demography of family life may threaten the availability of adult children to
provide help and support. Fertility rates have been declining since the “baby
boom” cohort and are projected to stay low (Siegel, 1993). With fewer children
per family, the elderly will have fewer children to draw on for support (Crimmins
& Ingegneri, 1990). Furthermore, it has been suggested that increasing divorce
rates and increasing rates of labor force participation among women may affect
the extent to which families can and will provide support, and increasing
migration patterns could lead to scattered families, further weakening the
effectiveness of the family as a support network (Crimmins & Ingegneri, 1990).
These demographic changes, thus, create an issue of kin availability (Easterlin,
Schaeffer, & Macunovich, 1993).
Social changes, therefore, lead on the one hand, to the increase in
parental caregiving needs, and on the other hand, to a possible weakening of
kin availability. In such climate, it is imperative to seek a better understating
and to predict what influences adult children to provide help and support to their
aging parents. While early research (i.e., Cantor, 1983; Stone et al., 1987)
addressed this topic from a cross sectional point of view, examining the
constraints of the adult children at the time where help and support were
needed, recent research (i.e., Parrott & Bengtson, 1999; Silverstein, Parrott &
4
Bengtson, 1995; Whitbeck, Hoyt & Huck, 1994) has focused on the longitudinal
aspect of intergenerational relationships and the effects of earlier family
experiences on the exchange of help and support from adult children to their
aging parents at the time of needs.
This dissertation aims at contributing to a better understanding of the
complex and possibly reciprocal relationships between the familial, cultural, and
religious contexts and the degree to which adult children feel obligated and
actually provide support to their aging parents over time. More specifically, the
objectives of this dissertation are: (1) to examine the strength with which adult
children are expected to be responsible for the care of their aging parents, and
how this value changes across the lifespan, over several decades of historical
time, between successive generations, and in relation to one’s generational
position in one’s family; (2) to examine whether parents can influence the
amount of support they will be receiving in old age from their adult children by
affecting their children’s motivations, normative obligations, or actual supportive
behaviors throughout the life course; (3) to examine the patterns of congruence
or discrepancy between filial norms--what people believe one should do to
support one’s parents, and filial behavior—what people actually do to support
their aging parents; and (4) to investigate the role of religion and religiosity in
predicting such patterns of discrepancy or congruence between norms and
actions.
5
C. Conceptual Framework
This dissertation is guided by an ecological multi-level dyadic
intergenerational model, which provides an explanation of the ecology or the
context within which intergenerational relationship between an adult child and
an aging parent over the life course takes place. The dyadic model focuses
simultaneously on the adult child and the aging parent as a family dyad. It
proposes that the motivation of adult children to care for their aging parents,
and specifically normative filial obligations and attitudes toward eldercare, can
be better understood within the multiple contexts, or ecological environments,
within which they develop. Such contexts also play a significant role in the
enactment of these norms and other motivations into actual provision of
parental support when such support is needed. This model builds and expands
on a preliminary model, which the author previously developed and discussed
in detail with regards to explaining risk factors for elder abuse in family settings
(e.g., Gans & Schiamberg, 1997; Schiamberg & Gans, 2000; Schiamberg &
Gans, 2005).
The model is derived from two broader theoretical perspectives: the
human ecological perspective, in particular the Bronfenbrenner (1979, 1986,
1997) model, and the life course perspective (e.g., Bengtson & Allen, 1993;
Elder, 1992; Elder, 1994; Elder & Johnson, 2002). I will first briefly describe the
relevant aspects of these broader theoretical perspectives and will then discuss
the ecological dyadic model suggested here.
6
Human Ecological Perspective
The ecological perspective proposes that human development and
intrafamilial processes are affected by extrafamilial contexts or environments
within which this development and these processes take place
(Bronfenbrenner, 1997; Bubolz & Sontag, 1993). The ecological perspective
developed in attempting to explain child development and especially the
interrelationship between schools, families and child development within
multiple macrolevel contexts such as public policies. I believe that this model
may serve as an insightful model in the study of aging in general and the study
of intergenerational support in particular.
Bronfenbrenner (1979, 1986, 1997) proposed a specific nested model
whereby the developing person at the focus of the model is surrounded by four
levels of environments including: (a) the microsystem: the most intimate
environment where human development takes place. Typically this
environment represents the family; (b) the mesosystem: the relationship
between the family and other principal settings in which human development
occurs. In the specific case of aging parents, this system may include the
relationship between informal support providers and formal services; (c) the
exosystem: environments which are external to the developing person but still
affect him/her, because these are environments in which other family members
participate in. For example, in the case of aging parents, the adult child’s
workplace is an exosystem. While the aging parent may not participate in the
7
child’s workplace, policies(i.e., leave of absence, flexible time etc.) may affect
him/her; And (d) the macrosytem: broad policies, ideological values, cultural
norms and institutional patterns of a particular society. An additional
component of this theoretical perspective is the chronosystem, a concept of
chronological time, capturing the influence of change and continuity over time
and with age on the development of the focal person.
The Life Course Perspective
Similarly to the ecological perspective, the life course perspective
provides both a developmental and historical framework for the study of
intergenerational relationships (Hareven, 1996) and emphasizes the importance
of change over time and the various micro and macro contexts within which
intergenerational relations take place (Bengtson & Allen, 1993). In the study of
aging the life course perspective helps drawing our attention to the interaction
of demographic forces and cultural factors in shaping generational relations and
family processes over the life course (Hareven,1995) and their accumulated
effect later in life.
The life course perspective is guiding research using five main principals
(Elder & Johnson, 2002): (a) The principle of life span development: Individuals
continue to develop meaningfully throughout life. In addition, late life adaptation
and aging should be studied from a longitudinal perspective as early
experiences in childhood and adulthood affect changes in late adulthood and
8
aging; (b) The principle of agency: Individuals construct their lives through their
actions and choices within the opportunities and constraints of history and
social circumstances; (c) The principle of time and place: The life course of
individuals is embedded within and shaped by historical events, and the effect
of the same event may differ across different places; (d) The principle of timing:
The same events may affect individuals in different ways depending on when
they occur in the life course; (e) The principle of linked lives: People’s lives are
interdependent and as such are affected by one another.
One of the most relevant assumptions of the life course perspective to
this dissertation is the “multiple clock assumption” (Hareven, 1995) suggesting
that there are three temporal contexts that affect human development:
Ontogenic time, generational time, and historical time. Ontogenic time refers to
events in the biography of a person. It is indexed by chronological age or by
age periods or stages. In the case of the first paper in this dissertation, for
example, the concept of filial maturity—a developmental stage whereby adult
children view their parents as possibly needing their help—is on the ontogenic
temporal axis. Generational time refers to the position of the individuals in the
lineage of the family and to events at the family level. The death of both
parents and the transition of one to the first generation in a lineage, as
discussed in the first paper of this dissertation, is an example of generational
time. Historical time refers to the macro social dimension of time and to events
9
in the broad social context. The debate over the weakening of the family or
family decline over historical time falls under this temporal axis.
The Ecological Multi-level Dyadic Intergenerational Model
For the purpose of this dissertation, the original Bronfenbrenner (1979,
1986, 1997) model was modified to incorporate a dyadic perspective, focusing
simultaneously on factors affecting the adult child and the aging parent. Please
refer to Figure 1.1 for a graphic view of the model. While the original model
addressed one focal developing person, this revised model focuses on
developmental processes at both the adult child and aging parent level.
Utilizing the life course perspective terminology, the ontogenic development of
both the adult child and the aging parent are the co-focus of this dissertation. A
specific concept at the ontogenic time of the adult child—filial maturity—is
discussed in detail in the literature review and examined in paper 1 of this
dissertation. The concept of filial norms—the obligatory responsibility adult
children feel towards the care of aging parents is the center of all three papers
in this dissertation. Other factors at the adult child and the aging level that are
known through previous research (see literature review) to be affiliated with
either the development of filial norms, the enactment of such norms into actual
care, or parental care are examined. Including, for example: religious behavior
and religiosity, as discussed in both second and third papers of this
dissertations, marital status (widowhood for aging parents, and married for adult
10
Figure 1.1
Conceptual Framework: Ecological Dyadic Intergenerational Model for Understanding Factors
Affecting Parental Support
Macrosystem
National context Ethnicity Policy
Religion & Religiosity Culture Family values
Exosystem and Mesosystems
Adult Child: Aging Parent:
Financial resources Formal services
Employment status Financial resources
Social Networks
Microsystem
Chronosystem (chronological Time)
Parental
Support
Instrumental
&
Social
/Emotional
Adult Child:
Filial Maturity
Filial obligation/norms
Religious behavior
Religiosity
Education
Marital status (married)
Gender
Aging Parent:
Rising Needs:
Poor Health
Widowhood
Religious behavior
Religiosity
Education
Gender
Interaction: Aging parent-Adult child:
Solidarity Socialization
Demonstration Geographic proximity and contact
Change in lineage roles (death of both parents)
Competing demands (e.g., adult child having young children)
Exchange of support
11
children), education levels, gender, and rising needs in terms of poor health and
advanced age on the part of the aging parents.
The interrelationship and interactions between these two focal individuals
are then examined considering various familial factors, which represent the
generational time in the life course perspective terminology, or the microsystem
in the human ecological theory. At this level, various factors are examined
including the death of both parents and the transition into a new role in the
lineage as a first generation (sometimes referred to as the alpha generation).
This concept is discussed in detail in the first paper of this dissertation as it
suggests that this generational transition affects the development of filial norms.
The second paper of this dissertation focuses on several factors at the
miscrosystem and looks at them over time suggesting that parents may affect
the development of filial norms earlier in life through various pathways including
socialization, demonstration of eldercare, exchange of support and fostering of
strong solidarity and emotional bonding across the life course. Other factors at
this level include geographical proximity, contact, and competing demands in
terms of the adult children having young children at home, all shown to be
affiliated with parental caregiving in previous studies. Geographic proximity and
frequency of contact are examined in detail in the second paper of this
dissertation and are proposed to be mediating factors affecting the level of
support adult children provide their mothers. The third paper of this dissertation
12
addresses the role of geographic proximity and competing demands as possible
predictors of discrepancy or congruence between filial norms and filial behavior.
These interrelationships and specifically the development of filial norms
and their enactment into actual parental care are then examined in more distant
contexts. In a dyadic model, the mesosystem of one focal person is the
exosystem of the other focal person. In other words, factors that affect the
aging parent outside of the family such as work status or financial resources are
included in his/her mesosystem. These factors may be included in the aging
parent exosystem. Although the adult child does not directly participate in the
child’s work place or his/her financial resources, these factors may still affect
the aging parent. For example, a working child may provide less support and a
child with more financial resources may be able to purchase more services
instead of providing support. Similarly, factors such as use of formal services
and financial resources of the aging parent may still affect the adult child even
though he/she does not participate in them directly. For example, an aging
parent who can afford to purchase formal services may not need as much
informal support. The role of financial resources is taken into account in the
analyses in all three papers of my dissertation as a control variable. The life
course perspective concept of linked lives is useful for understanding this
interrelationship between the dyadic meso- and exo-systems as it proposes that
one’s life is intertwined with other individuals’ lives and thus is both affected by
and affects other individuals’ lives.
13
The model further suggests that factors at the macrosystem should be
considered. Specifically for this dissertation, I am investigating the role of family
values. These values are examined across historical time, using the life course
perspective terminology, as the debate over the decline of family values is
discussed in the literature review and tested directly in the first paper.
Additionally, the third paper examines the role of religion and religiosity as
influential contexts within which filial norms develop and are enacted into actual
supportive behavior.
The Role of the Ecological Multi-level Dyadic Intergenerational Model in this
Dissertation
Building on Bengtson’s view (Bengtson, Acock, Allen, Dilworth-
Anderson, & Klein, 2005) on the role of conceptual or theoretical model, I
believe that the model presented in this dissertation serves several important
roles in this dissertation. First, it provided a useful and organized framework for
integrating what is already known about the development of filial norms and the
contextual factors that affect the enactment of such norms into actual
supportive behavior. Please see the next chapter for literature review. It then
allowed me to predict what is not yet known and needs to be investigated. In
the first paper, I addressed filial norms as a dynamic factor that changes over
time and the development of such norms over time. This was the first analysis
of its sort on this topic. In addition, I specifically tested the hypothesis of family
14
decline in the specific area of obligation to parental care. In the second paper, I
addressed the motivations of parental caregiving and how such motivations
develop within families over time. In the third paper, I expand our knowledge on
the macrolevel contexts within which filial norms develop and are enacted into
actual support. I specifically addressed the role of religion, religiosity, ethnicity
and national context, all of which were not addressed in previous studies.
The ecological multi-level dyadic intergenerational model further
prescribed or at least guided me in the choice of research methods. For
example, in order to examine the role of multiple texts in a nested arrangement
as suggested by the model, I utilized a multi level modeling in my first paper. I
also utilized longitudinal analysis in both the first and the second paper in order
to capture the concept of the chronosystem in the ecological perspective terms,
or the change and continuity over ontogeneic, generational and historical time,
using the life course perspective terminology. The model further allowed me to
explain why this knowledge is important as discussed in the beginning of this
chapter and in each paper individually.
It is important to note that the ecological model as well as life course
perspective have been critiqued for being too complex suggesting that one
analysis cannot take into account all of the factors proposed by the model (e.g.,
Bubolz & Sontag, 1993). However, I believe that the complex model allows one
to theoretically consider all the factors and attempt to at least control for them, if
not directly address them, to the best of one’s ability in his/her study. Each of
15
the analyses presented in this dissertation addresses a smaller section of this
complex model, allowing me to piece together the complex jigsaw puzzle that is
presented by this framework and represents intergenerational relationships over
the life course and in later life.
D. Literature Review
Norms of Filial Responsibility
Norms of filial responsibility are an aspect of a broader concept of
familialism. While familialism represents attitudes about the centrality and the
importance of the family and values surrounding help and support behaviors
between various family members (Parrott & Bengtson, 1999), norms of filial
responsibility examine more specific attitudes toward help and support from
adult children to aging parents. Filial responsibility is defined as the amount of
obligation and responsibility a person feels adult children should have for
assisting aging parents (Cicirelli, 1988, 1990). More than an expectation of
one’s own behavior, norms of filial responsibility refer to the recognized duties
and obligations that define the social role of adult children with respect to their
aging parents. Although expressed norms are predictive of personal intentions
to provide support and the supportive behaviors themselves (Bromley &
Blieszner, 1997; Peek et al., 1998; Silverstein & Litwak, 1993), they are
conceptually distinct from both intentions and support (Stein et al., 1998). Stein
et al. (1998) specifically discussed a distinction between more generalized filial
norms, such as those discussed in this dissertation, and more specific filial
16
intentions, or felt obligations, towards the care of one’s own parents. As
generalized expectations reflecting underlying value orientations, filial norms
are relevant to people at all stages of the adult lifespan regardless of
generational position, including children who have no surviving parents, and
individuals who have no children.
However, at a practical level, filial norms may change in response to
personal circumstances that affect the ability to provide parental care (e.g.,
competing demands) or cause one to retroactively attribute supportive behavior
to internalized normative values (e.g., providing parental care). Guided by the
theory of cognitive dissonance, Finley, Roberts and Banahan (1988) suggest
that generalized filial expectations may be adjusted in an attempt to reconcile
the gap between the ideal and what is possible or actual.
The study of filial norms is based on the hypothesis that a history of
stronger filial responsibility of adult children may predict the extent to which
those adult children provide actual help and support to their aging parents at a
later point in time when such support is needed. Such association between
norms, values or attitudes and actual behavior is supported by a larger
theoretical model – the intergenerational solidarity model. This section will
briefly present the theoretical model as it relates to the topic at hand and will
review previous studies that addressed similar topics.
17
Norms of filial responsibility as part of the intergenerational solidarity model
Intergenerational solidarity (Bengtson, 1996; Parrott & Bengtson, 1999)
is a multidimensional approach representing dimensions along which family
cohesion or integration occurs (Parrott & Bengtson, 1999). These dimensions
include several constructs of solidarity including affectual, associational,
consensual, structural, normative, and functional (Lee, Netzer, & Coward,
1994). Several studies (e.g., Bengtson & Roberts, 1991; Parrott & Bengtson,
1999; Whitbeck et al., 1994) have examined interrelations among subsets of the
intergenerational solidarity dimensions and the empirical evidence confirms the
theory and demonstrates its utility in examining the nature and the quality of
relationships between parents and adult children (Allen, Blieszner, & Roberto,
2000). However, fewer studies had focused on the association suggested in
this paper - between normative and functional solidarity (Lee et al., 1994).
Normative solidarity refers to familialism and is defined as the strength of
commitment to perform familial roles and to meet familial obligations (Bengtson
& Roberts, 1991). Norms of filial responsibility represent an aspect of that
concept and are, as previously discussed, the amount of obligation and
responsibility adult children should have for assisting aging parents (Cicirelli,
1988, 1990). Functional solidarity refers to the extent to which help and support
is given, received, or exchanged between family members (Parrott & Bengtson,
1999).
18
Evidence for the relationship between normative solidarity (filial responsibility)
and functional solidarity (the actual provision of support)
Bromely & Blieszner (1997) suggested that the sense of filial
responsibility an adult child feels to provide support for his/her aging parent
affects the actual caring for that parent. In fact, there seems to be a clear
conceptual connection between the concepts of filial responsibility and
functional solidarity. The former involves norms related to the intergenerational
exchange, and the latter looks at the actual intergenerational exchange (Lee et
al., 1994). Previous studies found direct relationships between the strength of
filial norms held by adult children and actual provision of support. This
relationship held true over time when filial norms were measured earlier in time.
For example, Silverstein and Litwak (1993) provide empirical evidence
suggesting that stronger filial obligation increases the likelihood that adult
children will provide instrumental support. Silverstein et al. (1995) found that
among middle-aged sons, stronger endorsements of norms of familialism are
associated with greater support to aging parents. This statement does not hold
true for daughters who are influenced by intimacy and altruism, whereas sons
are influenced by normative principles, familiarity, and the expectation of
financial reward. Silverstein, Gans and Yang (2006) similarly found that filial
norms as expressed by adult children earlier in life are related to the actual
support provided to their parents later on in life. Whitbeck et al. (1994) found
that filial concern influenced contemporary intergenerational help and support
19
exchanges. Willingness to help and support a parent was, in part, mediated by
concern for the parent and a sense of duty.
Other researchers addressed the relationship between filial expectations
-- the extent to which parents are expecting their children to support them in old
age, and actual support provided to them by their children. Results from these
studies are less clear. Several studies provided little evidence to such
relationships. For example, Eggebeen and Davey (1998), using a sample of
the National Survey of Families and Households found that supportive behavior
on the part of children is not linked to parental expectations. Similarly, Lee et
al. (1994), in their investigation of the association between filial responsibility
expectations and patterns of intergenerational assistance found that aging
parent’s filial responsibility expectations are positively related to aid given to
their children (when parental resources such as education, income, and health
are controlled), yet not related to the amount of help provided by the adult
children. Building on that work and extending it, Peek et al. (1998) suggested
that parents might modify their social norms regarding receiving care from adult
children (global filial responsibility expectations from their adult children) based
on their children’s circumstances and resources. Those adjusted norms are
termed specific expectations or behavioral intentions to seek care. Using data
on impaired elderly parents who lived in the community, the authors found that
the adjusted expectations of care – the specific expectations or behavioral
intentions as adjusted based on the child’s reality, influence the amount of care
20
received from the adult children. In contrast, Lowenstein, Katz, and Daatland
(2004) in their study, titled Old Age and Autonomy: The Role of Service
Systems and Intergenerational Family Solidarity (OASIS), a nationally
representative cross national study of five countries, found a significant effect of
filial expectations held by aging parents on the amount of help provided to them
by their adult children in four of the five countries (Norway, England, Germany,
and Israel). Due to the stronger evidence to a relationship between filial norms
held by adult children and provision of support to parents, this dissertation
focuses on filial norms held by adult children and not on filial expectations held
by the parents.
Determinants of Filial Norms and Actual Supportive Behavior
Previous research indicated several significant predictors to the extent to
which support is given, received, or exchanged. Several of these predictors are
related to the adult children, others to the aging parent and yet others to both
the adult children and the aging parents and to the larger contexts within which
their relationship take place. Based on the conceptual model that guides this
dissertation, the literature review of the determinants of parental support is
organized into the following sections: Characteristics of the adult children,
characteristics of the aging parents, the microsystem: characteristics of the
interrelationship between the adult children and their aging parents, the meso-
and exo- systems of the adult child and the aging parents, and the predictors at
21
the macrosystem including race, ethnicity and national contest. The discussion
on the role of time is embedded within each of those sections.
Characteristics of the adult child
One of these predictors is gender of the adult child (Silverstein et al.,
1995). Findings regarding gender differences in filial norms toward older
parents are mixed (Stein et al., 1998). In general, women and men do not
seem to differ markedly in their expression of filial obligations (Connidis, 2001).
Although men and women are viewed to be equally obligated to provide support
to elderly family members (Ganong, Coleman, McDaniel, & Killian, 1998), there
is some evidence for gender differences with regard to specific types of support
provided. For example, Finch and Mason (1991), using a vignette technique in
a British sample, found that adult sons were expected to provide financial
support and adult daughters, personal care and housing. In the U.S.,
researchers (e.g., Ganong and Coleman, 1999; Hogan, Eggenbeen, & Clogg,
1993; Spitze & Logan, 1989) found similar results showing that sons are
expected to and are likely to provide instrumental help and financial assistance,
and daughters to prepare meals and clean. Daughters were also more likely to
provide advice.
Another predictor is income or more generally, social class, as it affects
both the need and the availability of support (Hirdes & Strain, 1995; Lee et al.,
1994). Financially and educationally advantaged families can more easily
22
purchase care in the private market, thereby diminishing their sense of
obligation to provide care themselves (Finley et al., 1988). Conversely, there is
a fair amount of consensus that working- and lower-class families hold stronger
filial obligations because of their need to be self-reliant (Connidis, 2001).
There is a vast body of literature on the impact of marital disruption (as
well as remarriage) on parental support and filial obligation. Divorce is viewed
as a disruptive factor in the support exchange between parents and children
(Davey, Janks & Salva, 2004). Some findings suggest that divorced children
feel less filial obligations than those who have not experienced a marital
disruption (e.g., Cicirelli, 1981). The divorced have less money and time to help
other family members (Ganong & Coleman, 1999), and may face challenges
that may make them less sensitive to the needs of their parents (Connidis,
2001).
An important factor to consider when addressing the adult child’s
characteristics is the possible view of filial norms as a developmental process
on the ontogenic time using the life course perspective terminology, as
proposed in the first paper of this dissertation.
Filial norms development on the ontogenic time: The concept of Filial maturity
The concept of filial maturity was proposed by Blenkner in 1965 and
defined as an individual developmental stage. Nydegger (1991) further
developed this term suggesting that it represents a developmental stage in
23
terms of the Ericksonian developmental tasks. This developmental stage is
heralded by the onset of a filial crisis, the resolution of which results in the
attainment of a developmental period. This period is characterized by a change
in perspective, which allows the adult child to take on “a filial role- a caregiving
role, which involves being dependable insofar as his parent is concerned”
(Blenkner, 1965, p. 57). The change in perspective refers to the ability of the
child to see his/her parent as an individual with his/her own rights and needs,
and therefore, to provide care in the context of adult-adult relationship, with a
possible role reversal. Blenkner’s (1965) discussion implies that there is an
assumption that filial maturity is an age specific stage that occurs as a midlife
recognition. Murray, Lowe and Horne (1995) pointed out the similarity of
Blenkner’s term to Carl Jung’s mid-life crisis that is expected to occur between
40 to 60 years of age.
Cicirelli (1988, 1990) contributed another term to the discussion – filial
anxiety. Filial anxiety is related to the concern that adult children have regarding
the amount of help they will be called upon to provide their aging parents in the
future, and whether they will be able to manage that burden. Although the
concept was originally suggested as relevant to adult children who were not
involved in caregiving, Cicirelli (1988) suggested that it could be extended to
situations where parental care is being provided. He suggested that “filial
anxiety could thus occur at a later point along the trajectory of decline, when the
adult child was already providing some care, but was worried about anticipated
24
further decline of the parent’s health, and whether additional needs of care
could be met” (Cicirelli, 1988, p.478). Bromley & Blieszner (1997), in their
research on planning for long-term care, found evidence that supports Cicirelli’s
(1988) thesis. In their research, the majority of adult children reported having
thought about and considered the future dependency needs of their
independent parents. Some of the adult children also were involved in some
intergenerational discussions of possible options (Bromley & Blieszner, 1997). It
seems, therefore, that issues associated with the increasing needs of an aging
parent may trigger filial anxiety. In further support of Cicirelli’s thesis, Murray et
al. (1995) confirmed that adult caregivers between the ages of forty and sixty
experienced high levels of filial anxiety, which could reflect the existence of a
filial crisis.
Although not yet suggested in previous studies, the first paper of this
dissertation proposes that if indeed one goes through a midlife filial crisis that is
characterized by high levels of filial anxiety, one should also be, at least
theoretically, expected to develop and exhibit higher levels of filial norms
following such crisis. Such higher norms seem to be related to the achievement
of filial maturity in preparation for assuming the required filial role of caregiving.
It is, therefore, reasonable to expect that there should be an association
between age and social norms. More specifically, it seems that norms should
be lower in young adulthood, peak at midlife and decline somewhat in old age.
25
Characteristics of the aging parent
Gender of the parent may play a role in the types of support received.
Mothers were found to receive more support than fathers (e.g., Ikkink, Tilburg,
& Knipscheer, 1999). Gender of the parents may interact with other factors
such as, for example, marital status of the parent. Parental divorce may
weaken the filial norms of children, especially toward divorced fathers who are
less likely to reside with their children after the divorce and more likely to
remarry than divorced mothers (Ganong & Coleman, 1999). As a result, fathers
who got divorced from mothers when the children were growing up and did not
have custody of their children may receive less support than the mothers in old
age. In addition, divorces and remarriages in the parental generation may
increase the number of potential care recipients per adult child by adding
stepparents and other step relatives. It is possible that having more kin as
potential care receivers would diminish the amount of care given to each one
(Ganong et al., 1998). Marital status of the parents play a significant role in
support received in another way, where widowed and divorced parents receive
more help than married parents (e.g., Ikkink et al., 1999; Rossi & Rossi, 1990).
Another predictor is income or more generally, social class, as it affects
the need for support. Aging parents with higher financial resources can
purchase services and may access services of higher quality, thereby possibly
diminishing the need for some types of informal care from their children (Hirdes
& Strain, 1995; Lee et al., 1994).
26
Health serves as a strong indicator of the need for assistance (Parrott &
Bengtson, 1999) where individuals with poor health status are much more likely
to require assistance than those with better health. This is particularly true
when examining instrumental support where the tasks often demand physical
strength and endurance (Hirdes & Strain, 1995). Thus, poor health of the aging
parent should predict increased exchange of help and support. Similarly,
advanced age of the aging parent was found to be a predictor of higher levels of
functional support (Hirdes & Strain, 1995; Morgan, Schuster, & Butler, 1991).
Microsystem: The interrelationship between the adult child and the aging
parents
Several studies (e.g., Lowenstein et al., 2004; Parrott & Bengtson, 1999;
Walker, Pratt. Shin, & Jones, 1990; Whitbeck et al., 1994) showed that affectual
solidarity between parents and their children was significantly affecting
exchange of intergenerational support. Walker et al. (1990) suggested that filial
obligation and affectual solidarity occur at the same time when adult children
reflect on future caregiving. Silverstein et al. (1995) further found that affectual
solidarity interacted with gender in its effect on provision of support.
Specifically, they found that for daughters, strength of emotional bonds between
them and their parents represent a direct motivation for parental caregiving,
whereas for sons, such emotional bonds influence parental support indirectly by
increasing contact with the parents.
27
Ribar and Wilhelm (2006) discuss the possible role of social learning or
demonstration of eldercare on parental caregiving, specifically attempting to test
whether parents’ support to their own parents provides a model to their children,
which the children are then willing to imitate when the time comes for them to
provide support to their parents. Results from previous studies on the role of
behavior modeling are mixed. Several empirical studies revealed evidence in
support of social learning theories. For example, using nationally
representative data from the National survey of Families and Household
(NSFH), Szinovacz (1997) found that a respondents’ attitudes about bringing
their aging parents to live with them was positively related to whether
grandparents lived with them in their parents’ house when they were growing
up. Using the three generations study of Mexican Americans, Ribar and
Wilhelm (2006) found that the attitudes towards coresidence with elderly
parents and towards the financial responsibility to aging parents of the youngest
generation were affected by help the middle generation provided to the oldest
generation. However, using a sample of women from the National Longitudinal
Surveys, Caputo (2005) suggested that parents’ caregiving behavior to their
own parents did not influence their daughters’ caregiving activity.
Another predictor addressed by previous literature is family structure.
The presence of siblings in the family is a predictor of intergenerational
assistance, especially that given to an aging parent. Only children are relied
upon for all necessary support while individuals with siblings can share
28
responsibilities (Parrott & Bengtson, 1999). Most research in the U.S. (e.g.,
Eggebeen, 1992; Uhlenberg & Cooney, 1990) shows that elderly parents who
have more children receive higher levels of support in total. However,
Lowenstein et al. (2004) in their study of four European countries (Norway,
England, Spain, and Germany) and Israel found that the number of adult
children in one’s family negatively affected the amount of care provided by each
of the children. They proposed that this finding may represent that the amount
of help is divided among the children resulting in less support given by each
child, but possibly more support given in total.
The presence of young children in the adult child’s house has been
extensively discussed as a possible source of competing demands that might
diminish capability of adult children and especially women to provide support to
their aging parents. Terms such as “women in the middle” (Brody, 1981) and
the “sandwich generation” were coined and vastly used in social gerontology
literature as metaphors for what seemed to be a frequent condition: women who
are “caught in the middle” between two important tasks: caring for children and
caring for aging parents. Early literature (e.g., Brody, 1981) described this
condition as detrimental to the well-being of middle-aged individuals as well as
possibly to their aging parents, who might not have their caring needs met by
their overburdened daughters (or sons). These reports were usually based on
qualitative work and anecdotes. Quantitative analyses from various studies did
not provide consistent evidence to the suggested negative effect of competing
29
demands in the form of care for young children on the amount of elder care.
While Hamon (1992) found that having more dependent children led to
decreased parental support, other studies (e.g., Himes, 1994; Loomis & Booth,
1995) suggested that the presence of young children did not affect the amount
of parental support. In addition, demographic studies suggested that the
likelihood of women being caught in the middle between care for young children
and care for aging parents is small (Himes, 1994).
Other determinants at the microsystem include geographic proximity and
frequency of contact. Evidence from various studies in the U.S. and elsewhere
(e.g., Hogan et al., 1993; Lowenstein et al., 2004; Rossi & Rossi, 1990)
demonstrate that greater geographic distance and limited face-to-face contact
reduces the incidence of all types of intergenerational exchanges and
represents a significant barrier to intergenerational support. Hogan et al.
(1993), however, showed that more modern forms of communication may
lessen the effect of physical distance at least for some forms of support.
Frequency of contact was further shown to play a role at least for sons in the
amount of support provided to their parents (Silverstein et al.,1995).
The mesosytem and exosystem
As previously described in chapter 1, in the dyadic model, the mesostem
of one focal individual may represent the exosystem of the other focal person.
Based on the Bronfenbrenner (1979, 1986, 1997) model, the mesosystem
30
includes the relationship between the focal person, his/her family and other
more distant institutions that directly affect him/her. The exosystem includes
other institutions that the focal person does not directly participate in but that
still affect his/her life. In the case of the dyadic model presented here, factors at
the mesosystem of the adult child such as the workplace, have an effect on the
aging parent and are thus in his/her exosystem. Similarly, factors at the aging
parent’s mesosystem, such as use of formal services, may still affect the adult
child who does not directly participate in them, and are thus in his/her
exosysetm.
There is a plethora of literature and studies discussing the effects of
employment status on parental care under the assumption that having a job
represents a competing demand to parental support for adult children. Casper
and Bianchi (2002) described family and work as two greedy institutions
competing for one’s time and energy. They suggests that balancing work and
family life, although not a new problem, is more challenging today as work life is
more demanding and will continue to be demanding in the future when the baby
boomers retire and their load of work would have to be absorbed by less
workers. Because women are more likely to provide support, most research
addressed the effects of women’s participation in the labor force on parental
support. However, most research showed that women’s paid work does not
change the likelihood of caring. For example, while Stoller (1983) found that
full-time employment significantly reduced the amount of care sons provided to
31
their parents, full-time employment did not have a significant effect on the
amount of assistance daughters provided. Spitze and Logan (1991) found no
effects of employment for men or women on the amount of contact by phone
with parents, feelings of closeness to parents or provision of assistance to
parents. Similarly, using a nationally representative sample that included
extensive measurements on both employment and parental support, Rosenthal
(2004) found that overall employment status did not significantly affect the
amount of time daughters and daughters-in-law spent on caregiving to elderly
parents and parents-in-law. However, unemployed daughters and daughters-
in-law spent more time on meal preparation, and women who are employed
part-time provide less emotional support to their parents or parents-in-law.
When faced with the need to provide parental caregiving, employed women
were found to use various strategies that allow them to provide that care
including: sacrifice their leisure activities and sleep; give up opportunities for
promotion; use vacation and sick days for parental care; cut back on some
housework activities; and reduce work hours (e.g., Connidis, 2001; Horowitz,
1985; Ikkink et al., 1999). It was suggested, however, that the competing
demands of employment and parental care may be met differently in lower
socioeconomic classes (Connidis, 2001). For example, Mutschler (1994), using
a sample of women who held mostly blue-collar and clerical jobs, found that
these women are likely to leave their jobs or at least reduce their work hours to
32
assume parental care. Connidis (2001) suggests that this may be true for
women who hold lower paying and lower-status jobs.
The availability of social relationships for aging parents may predict the
need for support from family members. Morgan et al. (1991) argue that
exchanges vary depending on the size of the social network. Individuals with
relatively few network members have fewer opportunities to make exchanges.
Moreover, individuals who live with others may have very different exchange
patterns than those who live alone. Living with another person, regardless of
the type of relationship, is conducive to receiving assistance (Stone et al.,
1987). In other words, not living alone is likely to be a more important
determinant of exchange than marital status per se (Hirdes & Strain, 1995).
The macrosystem
An important factor at this most general level of contextual variables is
the national context or the type of regime. This is specifically important in the
context of aging and intergenerational support because the type of regime may
affect the level of services available to older adults, which, may, in turn, affect
level of needs for informal support. The relationship between the welfare state
and families has long been in a center of debate (Motel-Klingebiel, Tesch-
Roemer, & von Kondratowitz, 2003). On one hand, economists have
suggested that services offered by the welfare state may “crowd out” families
and thus lead to less efficient service provision. This idea is also described in
the literature as the substitution hypothesis, suggesting that the provision of
33
services provides a substitution for families. On the other hand, researchers
suggested that families and formal services may complement each other by
providing services simultaneously. For example, Litwak, Silverstein, Bengtson
& Wilson Hirst (2003), guided by task-specific theory, suggest that formal
organizations are optimal in managing technical tasks in caregiving while
primary groups, such as families, are optimal for managing nontechnical ones.
As a result, they suggest that a partnership between formal organizations and
primary groups is the best approach to negotiate informal and formal care, with
each partner providing a different type of support. Finally, others further
suggested that the introduction or existence of formal services may encourage
further support from families. This hypothesis is the exact opposite of the
“crowding out” hypothesis and is sometimes referred to as the “crowding in”
hypothesis (e.g., Kunemund & Rein, 1999).
There is not much empirical support to suggest a process of “crowding
out” (Motel-Klingebiel et al., 2003). In fact, most evidence points to a process of
“crowding in” suggesting that elderly who reside in welfare regimes that provide
generous formal services are more likely to receive support from their children
(e.g., Kunemund & Rein, 1999). In a more recent paper using data from the
Old Age and Autonomy: The Role of Service Systems and Inter-generational
Family Solidarity (OASIS) research project of five countries (Norway, England,
Germany, Spain and Israel), Motel-Klingebiel, Tesch-Roemer, & von
Kondratowitz (2005) report that the total quantity of help received by older
34
people is greater in welfare states with a strong infrastructure of formal
services.
Another area that received much attention in the literature in the United
States is the role of race and ethnicity in determining both filial norms and
parental support. Traditionally, Black culture as well as Mexican-American and
other Hispanic cultures were viewed as more familistic than the White culture
(Burr & Mutchler, 1999; Davey et al., 2004; Hogan et al., 1993), suggesting that
such Black and Hispanic families may respond better to the needs of their
elders (e.g., Taylor, Chatters & Jacskon, 1993). In the case of the Black
culture, this tradition was based on the assumption that Blacks have suffered a
long history of discrimination that led to a decreased access to economic
resources and formal services as well as stronger reliance on family support
(Lee et al., 1994). In the case of Hispanic families, researchers suggested that
immigration patterns may have created stronger traditional familistic attitudes
(Burr & Mutchler, 1999). Several studies consistently showed that Black and
Hispanic elders have stronger filial norms than their White counterparts even
when other factors such as socioeconomic and actual receipt of formal support
and services are controlled. For example, Lee et al. (1994) reported that Black
older adults had significantly higher filial expectations of receiving care from
their children than did their White counterparts. Burr & Mutchler (1999) found
that both Black and Hispanic elderly had higher levels of endorsement for
intergenerational coresidence and financial assistance than their White
35
counterparts. However, empirical evidence does not always provide support to
suggest that these stronger norms are translated into higher levels of actual
support. Some studies showed some support to higher levels of support. For
example, Laditka and Laditka (2001) found that Black daughters provided more
help to parents than did White daughters. In addition, Shuey and Hardy (2003)
found that non-Whites were 3 times more likely to give financial assistance to
parents than Whites. However, several studies (e.g., Hogan, Eggenbeen, &
Clogg, 1993; Silverstein & Waite, 1993) found that Blacks received the same or
even lower levels of informal support than Whites. Similarly, Hogan et
al.,(1993) found that Hispanic families did not provide more support to their
elderly than did White families. Hogan et al. (1993) attempted to explain the
discrepancy between higher levels of familism in minority cultures yet similar or
lower levels of actual support. They suggested that the migration experiences
of some populations, such as some Hispanic groups, may lead to greater
geographic distance, which, in turn, leads to reduced frequency of contact. As
a result, even though these cultures may have stronger family norms, greater
geographic distance and reduced contact present constrains that impede the
process of translating these norms into action. Sarkisian and Gerstel (2004),
using the National Survey of Families and Households, suggested that it is not
necessarily the level of support that differentiate Blacks from Whites, but rather
the type of support. Specifically, they found that Blacks were more involved in
instrumental types of support such as transportation, household chores etc…,
36
while Whites were more involved in financial and emotional types of support.
Interestingly, these differences were mostly between Black and White women,
while men were quite similar.
Other studies addressed the cross-cultural and cross-national variations
in filial norms and intergenerational support. Most notably, Lowenstein and
Daatland (2006) investigated this topic using the cross-national five country
study—Old Age and Autonomy: The Role of Service Systems and
Intergenerational Family Solidarity (OASIS) study. Their results show that
overall in all countries (Norway, England, Germany, Spain, and Israel) the
respondents endorse the idea of filial norms and obligations. However, the
norms were endorsed more strongly in Spain and Israel than they were in the
other three countries with England and Norway showing as least familistic of all
five nations. Lowenstein and Daatland (2006) suggest that at least partially
these findings may be explained by the national context and specifically the
type of welfare regime in the country and the existence of lack of legal
obligation to provide financial support to parents, as well as cultural variations
such as the culture of familism in each country. What seems to be missing from
the literature is analyses of the role religion as well as religiosity may play in
affecting both filial norms as well as actual patterns of support. Among other
topics, this dissertation aims at adding this to the existing body of literature.
An important factor to consider at the macrosystem level is the concept
of historical time or continuity and change over time in our society as suggested
37
by both the life course perspective and the ecological perspective. Specifically
to this paper, it is important to address the debate over the change in familism
in general, and specifically in filial norms over time. The following section will
highlight this debate.
Family change or family decline: Is there a weakening of familistic and filial
norms?
While family scholars agree that families have undergone change over
the past century, they are not in agreement with regards to the nature and the
consequences of that change. In the late 80’s, David Popenoe (Popenoe, 1988)
presented a pessimistic view of family decline due to individualization. His
hypothesis of family decline led to a heated scholarly dispute that took place in
the early 90s, with some scholars joining his pessimistic view and others (i.e.,
Kain, 1990) calling it “the myth of family decline”. Scholars from both sides of
the debate base their propositions on empirical evidence. With regards to
familistic norms, there is much evidence that point to the resilience of family
values. For example, Bengtson (2001), using the same sample as this
dissertation (LSOG), reported a consistent result of high levels of affectual
solidarity over 26 years, reflecting consistent high levels of intergenerational
emotional bonds. Daatland and Herlofsson (2003), using the data of OASIS, a
comparative study of five countries (Spain, Israel, Germany, England, and
Norway) reported in their study on normative solidarity, that filial norms are
resilient providing evidence to what they termed ‘changed solidarity’ rather than
38
‘lost solidarity’. In the first paper of this dissertation, this hypothesis is put to the
test.
39
Chapter Two: Data Description: The Longitudinal Study of
Generations (LSOG)
A. General Description of the Sample
Data from the University of Southern California Longitudinal Study of
Generations (LSOG) are used for the analyses presented in this dissertation.
The LSOG began in 1970-1971 as a cross-sectional mailed survey titled the
Study of Generations and Mental Health, funded by the National Institute of
Mental Health (NIMH). The original sample included 2,044 respondents ages 16
to 91 from 328 three-generation families. The sample was selected via a
multistage stratified random sampling procedure from a population of 840,000
individuals enrolled in Southern California's first large Health Maintenance
Organization (HMO) – Kaiser Medical Group. All available grandparents (G1),
parents (G2), and grandchildren 16 years of age or older (G3) in the selected
families were eligible for the 1970-1971 study. The strategy was to select the
G1 members who had children and grandchildren between the ages of 16 and
26. The HMO primarily served union members in the steel industry and as a
result the original sample consisted of mostly white working-and middle-class
families. Minority families were underrepresented, yet education levels were
comparable with national norms (Bengtson, Biblarz, & Roberts, 2002). The
study involved assessments of each family member’s social attitudes and
values, physical and mental health, educational and occupational status, and
40
relationships with other family members including spouses as well as multiple
generations both up and down the lineage ladder.
In 1985, 1,331 of the original sample members were surveyed again;
Since then data have been collected at about three-year intervals (1985, 1988,
1991, 1994, 1997, 2000-2001; and 2005). The longitudinal response rate
between 1971 and 1985 was 73%, and has averaged 80% between waves
since then, a rate that is comparable to most long-term longitudinal surveys.
The sample is continually replenished by the addition of newly eligible spouses
and previously nonresponsive sample members (Parrott & Bengtson, 1999).
B. Description of the Unique Features of the Dataset
Several features of the LSOG, as described by Bengtson et al. (2002),
make it a particularly suitable data set in addressing the various research
questions posed by this dissertation. First, the LSOG has a longitudinal design.
The LSOG provides repeated assessments of all parts of the survey allowing a
true longitudinal study. The longitudinal design allows investigating change
overtime and across the lifespan, following individuals over the majority of their
adult lifespan. The first two papers in this dissertation take advantage of this
important feature. The first paper investigates change over four waves of data
(1985 - 2000). The third paper investigates lagged effects from an earlier point
in time (1971-1985) to a later point in time (2000).
41
Second, the LSOG provides a life course developmental overlap. The
study design of the LSOG created generations that have enough significant
overlap in age between them. As a result, researchers can compare younger
generations to the generation of their parents and even grandparents when they
were of similar age to them. Such generation-sequential design allows for
disentangling age-period-cohort effects. The first paper of this dissertation
takes advantage of this feature. Specifically, it addresses the developmental
growth or change in normative obligations over the lifespan—age effects, while
comparing the younger generation (G3) with their parent generation (G2)—
cohort effects, and over historical time between the 80s and the 90s-- period
effects.
Third, the LSOG has a multigenerational scope, looking at three and four
generations within families. Such design allows comparisons of
intergenerational relationships between various family members. Combined
with the longitudinal feature, this multigenerational scope allows investigating
continuity and change in the quality of such dyadic (i.e, parents and children of
various generations) or triadic (i.e., grandparents-parents-grandchildren of
various generations) intergenerational relationships over time. The second
paper in this investigation takes advantage of this feature by constructing
mother-child (G2-G3) dyads and examining various aspects of the mother-child
relationships over time. Given the longitudinal design, the second paper further
42
examines how various aspects of the mother-child relationship in the past affect
and shape present dyadic dynamics.
Fourth, the LSOG focuses on a unique in-depth set of psycho-social
measurements and measurements of family relations. This is a unique
combination of variables that allows researchers to investigate not only
individual change but change in family relationships over time. In addition, the
LSOG assesses social values and attitudes, allowing researchers to further look
at these relationships within the larger historical context. The first paper of this
dissertation takes advantage of this plethora of variables to look at the
contextual effects of familial and social factors on the developmental changes of
normative obligations over the lifespan. The third paper of this analysis looks at
a whole different aspect of family relationships, focusing on the role of religion
and religiosity on filial commitment and filial behavior.
The fifth unique feature of the LSOG is the combination of retrospective
and prospective measurements. Such combination allows investigation of more
than one point of view. For example, in the second paper of this dissertation, I
am using retrospective memories of the parents (G2) regarding providing care
for their own parents. In addition, I am combining these retrospective reports
with actual reports of support provision to parents from previous time waves.
Such combinations may reduce the chance of the various reporting errors and
biases attributed to each of these research techniques, improving the internal
validity of the study.
43
It is important to note that, this dataset is not without its limitations.
Given its original sample, the LSOG was mainly a sample of working-and
middle-class families, which were enrolled at Kaiser in 1970-1971. It therefore
excludes families that belong to the highest and lowest socioeconomic classes.
Additionally, individuals with lower educational backgrounds are under-
represented in the sample (Bengtson et al., 2002).
In addition, given the early date of the original data collection, the sample
does not adequately represent the ethnic and racial diversity typical to the
contemporary population of Southern California. The baseline sample in 1971
was 90% White. Even today, there are very few Black and Hispanic families as
well as Asian and other ethnic minorities, which are more predominant in
Southern California now than they were in 1971. The LSOG is not a nationally
representative data. As a result, conclusions from studies that are based on
this dataset cannot automatically be generalized to the general American
population. An extensive analysis by Bengtson et al. (2002) comparing the
1971 and 1997 waves of the LSOG data with census data from 1997 and the
1992-1993 wave of the National Study of Families and Households (NSFH),
suggested that the LSOG is comparable to national norms in its age, marital
status, and social class distributions. However, the LSOG clearly under-
represents ethnic minorities and the less educated. This may pose a problem
especially when discussing demographic trends such as religion affiliation and
religiosity, as discussed in the third paper of this dissertation, as well as
44
normative obligations and parental support that may be affected by cultural
traditions. The advantages and disadvantages of this dataset are further
addressed and discussed in the sixth chapter of this dissertation.
45
Chapter Three: Norms of Filial Responsibility for Aging Parents Across Time
and Generations
A. Introduction
Declines in mortality rates over the past century have increased the need
of older adults for prolonged periods of care, thus making caregiving to older
parents a normative activity in the lives of adult children (Brody, 1981). The role
of adult children in providing long-term support and care to their aging parents
has aroused much interest in social gerontology and family studies in the last
quarter century (e.g., Brody & Brody, 1989; Logan & Spitze, 1995; Silverstein &
Parrott, 2001), yet the norms of filial obligation that guide these supportive
behaviors are not well understood. The purpose of this investigation was to
examine developmental, historical, and generational dynamics in the
endorsement of filial norms toward older adults. Specifically, I examined the
strength with which adult children are expected to be responsible for the care of
their aging parents, and how this value changes across the lifespan, over
several decades of historical time, between successive generations, and in
relation to one’s generational position in one’s family.
Filial responsibility for assisting older parents is an aspect of the broader
concept of norms of familism--attitudes about the centrality or primacy of family
life (Parrott & Bengtson, 1999). As a social norm, filial responsibility reflects the
generalized expectation that children should support their older parents at times
of need (Cicirelli, 1988, 1990). More than an expectation of one’s own
46
behavior, norms of filial responsibility refer to the recognized duties and
obligations that define the social role of adult children with respect to their aging
parents. Although expressed norms are predictive of personal intentions to
provide support and the supportive behaviors themselves (Bromley & Blieszner,
1997; Peek et al., 1998; Silverstein & Litwak, 1993), they are conceptually
distinct from both intentions and support (Stein et al., 1998). As generalized
expectations reflecting underlying value orientations, filial norms are relevant to
people at all stages of the adult lifespan regardless of generational position,
including children who have no surviving parents, and individuals who have no
children.
At a practical level, however, filial norms may change in response to
personal circumstances that affect the ability to provide parental care (e.g.,
competing demands) or cause one to retroactively attribute supportive behavior
to internalized normative values (e.g., providing parental care). Guided by the
theory of cognitive dissonance, Finley et al. (1988) suggest that generalized
filial expectations may be adjusted in an attempt to reconcile the gap between
the ideal and what is possible or actual. Thus, in this investigation, I
acknowledge and try to capture the subtle distinction between structured and
situational aspects of filial norms as they evolve over the life course.
47
Intraindividual Change in Filial Norms
The literature on filial responsibility as a dynamic process falls along
three main lines of thought: (a) lifespan development, (b) cohort socialization,
and (c) life course. Lifespan developmental theories concerning filial
responsibility are typically focused on the psychosocial adjustments made by
individuals to meet family demands at successive life stages. With regard to
parent-care responsibility, Blenkner (1965) invoked the concept of filial maturity
to describe the transition of adult children from being relatively autonomous
from their parents to being dependable sources of support to them. This
transition involves a change in perspective that allows middle-aged children to
view their parents as vulnerable individuals, thereby strengthening their
commitment to provide care to them in the context of an adult relationship.
Murray et al. (1995) pointed out the similarity of Blenkner’s use of filial maturity
to Jung’s concept of the midlife crisis, and to Nydegger’s (1991) discussion of
filial maturity, in the Ericksonian tradition, as an aspect of personal growth that
follows from successful resolution of a family crisis.
Cicirelli (1988, 1990) proposed an alternative developmental framework
for conceptualizing filial responsibility based on caregiving anxiety. In his
framework, filial responsibility is induced when children worry about how they
might successfully manage care duties well in advance of the time that care is
actually needed by their parents. Anxiety over providing care to parents can
also occur at a later point in the caregiving process, when adult children worry
48
about the implications of further declines in their parents’ health (Murray et al.,
1995).
Filial anxiety and filial maturity may be mutually reinforcing
characteristics, as suggested by Bromley and Blieszner’s (1997) finding that
adult children who worried about the future dependency needs of their parents
also collaboratively discussed possible care options with them. Both
perspectives predict a heightened sense of responsibility on the part of adult
children in midlife when concerns for older parents are most likely to peak.
Studies have found little empirical evidence of midlife exceptionalism with
regard to filial duty to older parents, however, and, to the contrary, found a
linear decline in filial responsibility across successive age groups (Guberman,
2003; Peek et al., 1998; Rossi & Rossi, 1990). As these studies rely on cross-
sectional data, the nature of the developmental process remains unresolved;
without the ability to also consider cohort and historical differences among
different age strata, the attribution of such patterns to aging is only speculative.
Another perspective in the study of attitudes about family life is the
cohort-socialization model that focuses on stability and malleability of value
orientations. The most commonly found model under this rubric suggests that
central values are shaped by early socialization experiences in late
adolescence or early adulthood, and are unlikely to change in middle age and
beyond (i.e., Krosnick & Alwin, 1989; Sears, 1981). Much of this literature
emphasizes how historical conditions uniquely and, sometimes permanently,
49
shape the values of birth cohorts at an early age--when they are
“impressionable youth” (Alwin, 1990; Mannheim, 1928/1952; Riley, 1973;
Ryder, 1965). Under the strict assumption of this model--that values are
acquired early and then persist thereafter--any observed change in the values
of a population would come about through cohort turnover; that is, the
emergence of a cohort with value orientations different from the cohort it is
replacing. Applied to the present example, attitudes of a cohort toward parent-
care would be shaped by early exposure to culturally appropriate models of filial
duty based on the historical contingencies it encounters. If today’s older adults
were brought up during a period when commitment to family was stressed (e.g.,
the Great Depression or World War II cohorts), then the inculcation of values of
filial duty would be maintained throughout life. The values of society would only
shift when cohorts who newly entered the adult population were less familistic
than the values of an older generation who exited the adult population through
death.
In addition to the stochastic impressionable youth model, other models in
this genre have been advanced. These include lifelong openness, where
values are potentially malleable at any stage of the adult lifespan, and declining
stability, where change is more likely to occur at later stages of the lifespan
(Alwin & McCammon, 2003). As a general rule, however, this family of models,
unlike the normative approaches of many developmental models, remains
neutral about the direction and pattern of change that would be expected.
50
The life course perspective represents an ambitious attempt to
synthesize developmental, cohort, and social-structural approaches to the
understanding of change in attitudes over time (Elder, 1992). In the life course
framework, human development lies in the space where personal biography
intersects with historical events and social milieu. That is, trajectories of human
development are conditioned by the historical period and social context within
which development takes place. With regard to historical periodicity of attitudes
toward filial responsibility, much has been written about the declining salience
of family relations in American society. Social scientists, historians, and cultural
critics have noted that values of collectivism—including those of familism—have
been superseded by values of individualism in both civic and family life over the
last half century (Bellah, Madsen, Sullivan, Swidler, & Tipton, 1985; Hareven,
1996; Putnam, 1995; Roberts & Bengtson, 1999). Some scholars posit that
radical changes in family structure resulting from historically unprecedented
divorce and remarriage rates since the 1960s may have weakened the capacity
and willingness of adult children to provide support to their aging parents
(Crimmins & Ingegneri, 1990; Popenoe, 1988). Yet, other scholars consider
families to be resilient social organizations that are still capable of serving the
needs of their vulnerable members (Bengtson, 2001).
The life course perspective also considers social-ecological spheres that
influence human development, including such contexts as family, community,
region, and nation (Bengtson & Allen, 1993; Bronfenbrenner, 1979; Elder,
51
1992). Within families--the most intimate and fundamental of social
environments--human development is to a substantial degree bounded by the
interdependence of its members (Elder, 1994). Family units themselves pass
through stages of development along a temporal axis known as family time or
generational time that may be asynchronous with chronological age (Bengtson
& Allen, 1993; Hareven, 1996). For instance, adult children who experience the
death of their parents rise to become the oldest generation in their family
lineages, a succession that alters the allocation of roles and responsibilities
within the family system.
Interindividual Differences in Filial Responsibility
Research has identified social characteristics that differentiate individuals
by the strength with which their filial norms are held, including gender,
socioeconomic position, marital status (both of the adult child and the aging
parent), and the presence of young children in the household. Although
detailed literature review on each of these factors is beyond the scope of this
paper, I present here a summary of the main points based on the current
literature. I purposefully leave race and ethnicity out of this review because the
homogeneity of the sample used in the current research precludes their
consideration. I acknowledge, however, a rich literature that distinguishes the
cultural dimension of ethnic and racial background in shaping filial obligations
(for a review: Lee, Peek, & Coward, 1998).
52
Findings regarding gender differences in filial norms toward older parents
are mixed (Stein et al., 1998). Although, in general, women and men do not
seem to differ markedly in their expression of filial obligations (Connidis, 2001),
there is some evidence for gender differences with regard to specific types of
support. For example, Finch and Mason (1991), using a vignette technique in a
British sample, found that adult sons were expected to provide financial support
and adult daughters, personal care and housing. In the U.S., Ganong and
Coleman (1999) provided similar evidence, with sons expected to provide
instrumental help, and daughters to prepare meals and clean.
Socioeconomic status also affects the level of commitment to filial norms.
Financially and educationally advantaged families can more easily purchase
care in the private market, thereby diminishing their sense of obligation to
provide care themselves (Finley et al., 1988). Conversely, there is a fair
amount of consensus that working- and lower-class families hold stronger filial
obligations because of their need to be self-reliant (Connidis, 2001).
There is a vast body of literature on the impact of marital disruption (as
well as remarriage) on parental support and filial obligation. Some findings
suggest that divorced children feel less filial obligations than those who have
not experienced a marital disruption (e.g., Cicirelli, 1983). The divorced have
less money and time to help other family members (Ganong & Coleman, 1999)
and face challenges that may make them less sensitive to the needs of their
parents (Connidis, 2001). In addition, parental divorce may weaken the filial
53
norms of children, especially toward divorced fathers who are less likely to
reside with their children and more likely to remarry than divorced mothers
(Ganong & Coleman, 1999). Divorce of either generation may lower familism in
general.
Cognitive dissonance and attribution theories help explain why
individuals who have provided support to their aging parents develop a stronger
sense of filial obligation than those who have never provided such support (e.g.,
Bromley & Blieszner, 1997; Peek et al., 1998; Rossi & Rossi, 1990).
Strengthening filial norms is means of bringing attitudes and beliefs about
oneself into conformity with how one behaves (Finley et al., 1988). Conversely,
when support to older parents cannot be delivered, for instance because of
competing obligations, filial responsibility may be lowered (Bromley & Blieszner,
1997).
In summary, a variety of dynamic theories and approaches inform this
empirical investigation of filial norms about care and support to aging parents
over time. Although the perspectives reviewed are not mutually exclusive, each
emphasizes unique aspects of human development as applied to families. I
therefore, use these perspectives as guideposts to develop a more inclusive
framework for understanding if and how filial responsibility for older parents
changes over biographical, historical, and family time. Using data collected
from a unique longitudinal study of multigenerational families, I explore how the
strength of eldercare norms ebbs and flows along these various temporal axes.
54
B. Research Questions and Expectations
I begin with the general question of whether norms toward filial
responsibility are fixed dimensions of one’s personality or are pliable in
response to aging, life events, and changing historical trends. Drawing on the
filial maturity/anxiety perspectives, we anticipate an increase in the strength of
filial responsibility from young adulthood until middle age, followed by a
moderating tendency, or even a weakening in strength into old age. Other
dynamic forces likely to be correlated with the effects of aging are also
considered in our analyses: experiencing the death of both parents and
historical trends in filial responsibility to older adults.
I expect that generational succession, as a reflection of family time, will
alter the point of view of respondents with regard to filial responsibility toward
aging parents. Thus, I consider whether the occurrence of parental death is a
dynamic life event that signals a reappraisal of these norms. It is not possible
to determine whether parental death weakens norms by releasing altruistic
tendencies toward one’s own children (as one shifts perspectives from potential
provider to potential recipient of filial support), or rendering irrelevant one’s
parent-care duties; both mechanisms point in the same direction of change.
As I have discussed, contemporary accounts of family life have generally
pointed to a decline in the commitment to family over the last quarter of the 20
th
century. To address the issue of secular change in filial responsibility to older
55
parents, I directly test for period effects by assessing the historical time of
measurement as a dynamic factor in our models.
Several between-subject factors are considered as well that are
proposed to influence the level of filial norms and to modify the pattern of age-
related change. One of these factors relates to differences between
generations who were socialized in different historical periods. I anticipate that
the generation raised earlier in time (born in the 1920s and 1930s) will more
strongly endorse norms of filial support for older adults than will the generation
raised during a later period (born in the 1940s, 1950s, and 1960s). Other
between-subject factors included variables that have been found to be related
to filial commitment, including gender, education, marital status of the adult
child and the parent, having young children under the age of 18, and whether
one has provided help for one’s aging parents.
Finally, I expect that individuals related to each other by kinship (parent-
child, sibling, cousin, aunt/uncle-niece/nephew relations) would tend to share
similar beliefs. Family aggregations form an ecological context within which
common environments and genes may similarly affect the development of
individuals. Thus, I investigate the degree to which family members have
similar levels and trajectories of filial norms toward care of aging parents.
In this investigation, I thus ask the following research questions with
regard to norms of filial responsibility toward care of older individuals: (a) What
56
is the modal trajectory that underlies the expression of filial norms over the adult
lifespan? (b) What independent effect does the passage of historical time have
on the expression of filial norms? (c) Does the experience of generational
succession influence the strength with which filial norms are expressed? (d)
How do between-subject differences related to generation, gender, education,
the experience of family disruption (of both the adult child and the parent),
competing demands, and provision of parental support affect the expression of
filial norms, and (e) How similar are filial norms and their lifespan trajectories for
individuals within the same family?
C. Method
Sample
Data from the University of Southern California Longitudinal Study of
Generations (LSOG) are used for this analysis. The LSOG began in 1971 as a
mailed survey with an original sample of 2,044 respondents ages 16 to 91 from
328 three-generation families who were selected via a multistage stratified
random sampling procedure from a population of 840,000 individuals enrolled in
southern California's first large HMO. All available grandparents (G1), parents
(G2), and grandchildren 16 years of age or older (G3) in the selected families
were eligible for the 1971 study. In 1985, 1,331 of the original sample members
were surveyed again; since then data have been collected at three-year
intervals up to 2000. The longitudinal response rate between 1971 and 1985
57
was 73%, and has averaged 80% between waves since then, a rate that is
comparable to most long-term longitudinal surveys. The sample is continually
replenished by the addition of newly eligible spouses and previously
nonresponsive sample members (see Parrott & Bengtson, 1999, for details).
The sample reflects a diversity of social class backgrounds ranging from
working class to upper-middle class, but under-represents minorities because of
the source of the sample and the time in history when the families were
originally recruited.
The outcome variable for this analysis—filial norms—was measured at
four waves of data collection: 1985, 1994, 1997, and 2000. The universe of
eligible respondents in G2 and G3 generations was 2,359 of which 1,643 (70%)
responded to at least one survey. After omitting 16 respondents (1%) because
of item nonresponse on all surveys, the subsample used for this analysis
comprised 1,627 respondents from G2 and G3 generations who participated in
at least one of the four waves. I excluded the G1 generation because little
longitudinal data were available given their high rates of mortality over the
period, and the G4 generation because, as adolescents entered into the study
at later waves, they had not achieved the age necessary to make cohort-based
comparisons or to be at risk of family events of interest. These respondents
yielded 4,527 person-observations over the four waves of data, representing a
capture rate of 69%. The average age across all person-observations was 52.5
years (see Table 3.1).
58
Table 3.1
Age and Filial Norms at Four Times of Measurement (Observations = 4,527)
Age Filial Norms Year of
Measurement
n %
M SD M SD
1985 1,084 23.9 45.29 12.95 15.16 3.89
1994 1,177 26.0 52.44 12.74 14.08 4.46
1997 1,098 24.3 54.90 12.41 14.23 4.75
2000 1,168 25.8 57.02 12.24 13.94 4.67
Total 4,527 100.0 52.51 13.32 14.34 4.48
In terms of number of measurement per respondent, 37.6% contributed
four measurements, 25.2% three measurements, 15.3% two measurements,
and 22.0% one measurement. Because these percentages do not illustrate the
extent of sample attrition (because of sample replenishment, respondents with
fewer than four measurements may not have had the opportunity to participate
in all surveys), we calculated an attrition rate of 16.3% based on the number of
respondents who terminally exited the sample after their first participation in the
study.
A critical question in evaluating sample selection bias is whether
respondents who selectively drop out of the study (because of death,
incapacity, refusal, or loss to follow up) have characteristics that are associated
with study outcomes (Heckman, 1976). In this investigation, attrition bias would
59
be problematic if respondents with incomplete data had different intensities or
trajectories of filial norms from those who remained in the sample. In earlier
diagnostics (not reported), I found no significant differences based on attrition
status (comparing those who exited the sample to those who had complete data
after their year of study entry). Further analyses found no differences based on
number of responses. These diagnostics suggest that partial data are sufficient
to capture both levels and growth patterns of norms over time.
Another issue to consider with regard to possible attrition bias is whether
important predictor variables are related to the propensity of attrition. As
expected, I found that older, lower educated, and male respondents tended to
have fewer data points than their counterparts. I included factors that are
known to be related to attrition in the model as predictors. Controlling for these
variables in a latent growth model mitigates, but by no means eliminates,
attrition bias by adjusting for systematic biases that are associated with these
characteristics among those with incomplete data (see McArdle et al., 2004, for
a discussion of this assumption). Indeed, I specifically chose a multilevel latent
growth modeling approach because such a model is more flexible in terms of its
requirement for complete data, and treats repeated measurements as variably
nested within individuals and does not require that each individual in the sample
have a complete set of measurements, as is the case with traditional
multivariate repeated measures designs (Bryk & Raudenbush, 1992).
60
Measures
Dependent Variable. Normative filial responsibility for aging parents was
measured with the following question: “Regardless of the sacrifices involved,
how much responsibility should adult children with families of their own have:
(a) To provide companionship or spend time with elderly parents who are in
need? (b) To help with household chores and repairs and/or to provide
transportation for elderly parents who are in need? (c) To listen to the problems
and concerns of elderly parents and to provide advice and guidance? (d) To
provide for personal and health care needs of the elderly parent? (e) To provide
financial support and /or assist in financial and legal affairs of elderly parents
who are in need? (f) To provide housing for the elderly parents who are in
need?” For each item the respondent assigned responsibility on a five-point
Likert scale ranging from none to total. Because reliability coefficients for the
items were acceptably high in each year (.88 in 1985, .91 in 1994, .92 in 1997,
and .91 in 2000), responses were summed to create additive scales with a
potential range from 0 (no filial responsibility) to 24 (total filial responsibility).
Independent Variables. Predictor variables are grouped based on the
three levels of analysis: within-subject, between subjects, and family context.
Within-subject. Age (in years) at each wave of measurement was used to
predict developmental trajectories in norms at the level of the subject. Age
values are mean-centered at 52.5, and then divided by 10 to improve the
61
scaling of coefficients. A quadratic term of age-squared was included as well to
capture nonlinearity in temporal trajectories. Historical period was treated as a
dummy variable signifying whether responses were obtained from the 1985
wave of measurement versus the 1994, 1997, and 2000 waves. Parental
mortality was considered at each wave of measurement with a dummy variable
indicating whether both parents were deceased versus at least one parent was
still living. In addition, the presence of dependent children in the household was
represented as a dummy variable indicating whether the respondent had at
least one child under the age of 18 at the corresponding wave of measurement.
Distributions of these variables are shown in Table 3.2.
Between-subjects. All variables in this category were coded as dummy
variables. Generation membership was operationalized by comparing the later-
born generation (G3s) to the earlier-born generation (G2s). In 1985, G2s and
G3s averaged 57.6 and 33.1 years, respectively. Because these two groups
were identified by their lineage position not their birth year, they are arguably
too heterogeneous with respect to age to be considered cohorts.
Consequently, I refer to them as generations.
Sociodemographic predictors known to be associated with family outcomes
included gender (women vs. men), education (college educated vs. less
education), marital disruption of the adult child (experience of divorce vs. other
marital histories), and marital disruption of the family of origin—the parents—
during childhood. In the case of parental marital disruption, we did not have
62
Table 3.2.
Distribution of Explanatory Variables (N = 1,627)
Variables Number Percent
Level 1: Time Varying
Measurement by decade
1985 survey only 186 11.4
1994 - 2000 survey(s) only 543 33.4
Both 1985 and 1994 - 2000 surveys 898 55.2
Parent death/survival
Both parents deceased before 1985 381 23.4
Second parent died 1985 – 1993 128 7.9
Second parent died 1994 – 1996 48 3.0
Second parent died 1997 – 1999 61 3.7
Both parents alive in 2000 1,009 62.0
Child 18 years or younger in household
1985 survey 347 21.3
1994 survey 296 18.2
1997 survey 193 11.9
2000 survey 137 8.4
Level 2: Time Invariant
Gender
Female 906 55.7
Male 721 44.3
Education
College educated 607 37.3
Not college educated 1020 62.7
63
Table 3.2 continued
Marital disruption
Ever divorced 624 38.4
Never divorced 1003 61.6
Generation
G3 generation 955 58.7
G2 generation 672 41.3
Support provided to parents
Ever provided support 371 22.8
Never provided support 1256 77.2
Marital experience of parents
Parents divorced 295 18.1
Parents’ marriage intact 1014 62.4
Parents’ marital history unknown 318 19.5
information about many respondents because retrospective questions about
family structure during childhood were not asked in all surveys. To account for
the missing values, I created two dummy variables: one coded for those who
experienced parental divorce, and the other for those parental marriages of
unknown status (reference = parents’ marriage was intact). I also constructed a
behavioral support variable signifying that respondents provided their parents
assistance with household chores, transportation, or shopping at any
measurement point. The frequency distributions of independent variables are
presented in Table 3.2.
64
Family context. My interest in understanding the degree to which
common family membership jointly shapes norms leads to the consideration of
family context as the third level of analysis. I also note that because
respondents were recruited from the same nuclear and extended families, the
use of statistical techniques appropriate to such a nested data structure was
required. Thus, I estimated the extent of family-level homogeneity within the
333 families represented in the study. Because the sample followed entire
family lineages, the family groupings, or clusters, include siblings, cousins,
parents-children, as well as spouses and in-laws. The boundaries of family
membership are necessarily broad to accommodate the demands that
multilevel modeling imposes on the data. Family clusters averaged almost five
individuals per family, the largest family unit consisting of 19 individuals.
Statistical Approach
The nature of the data used to address the research questions required
a modeling procedure that could estimate time-varying, fixed, and group effects
in multiwave clustered data. Therefore, I used multilevel latent growth modeling
to test for age trends, historical change, and family-level variation in filial norms
toward aging parents (Bryk & Raudenbush, 1992; Raudenbush & Bryk, 2002).
The application of multilevel modeling specified three levels of analysis: within
persons (Level 1), across persons (Level 2), and within family (Level 3). The
analysis included 4,527 time-varying observations nested within 1,627
individuals, who were further nested within 333 families. To account for
65
incomplete data because of nonresponse I used full information maximum-
likelihood estimation to estimate parameters under the assumption that data are
missing at random or conditional on observed variables (e.g., McArdle et al.,
2004). This type of estimation offers a practical way to minimize attrition bias
and to increase the statistical power of estimated parameters by using all
available data.
At the first level of analysis, time-varying filial norm scores were
regressed on linear and quadratic age terms for each respondent across as
many as four measurements. These regressions generated random intercept
and slope estimates that described person-specific growth curves. Because
there are only four data points, some restrictions were necessary at Level 1 to
identify the model. I chose to estimate age-related coefficients as random
effects, as developmental change is our main focus. Other time-varying
coefficients are estimated as fixed effects, that is, as constant across individuals
and families. The equation at Level 1 is represented as:
y
ijk
= a
jk
+ b
1jk
t
ijk
+ b
2jk
t
2
ijk
+ b
3jk
p
ijk +
b
4jk
h
ijk
+
b
5jk
c
ijk +
e
ijk
,
where, y
ijk
is the measure of filial responsibility, t
ijk
is age (mean centered), t
2
ijk
is the square of age (mean centered), p
ijk
signifies whether both parents are
deceased, h
ijk
represents whether year of measurement was later than 1985,
and c
ijk
represents whether a child under 18 resided in the same household, for
the jth respondent in the kth family at the ith time of measurement. The
estimate a
jk
is the random intercept that represents the predicted level of filial
66
responsibility evaluated at the mean sample age of 52.5; b
1jk
is the random
linear slope for age; b
2jk
is the random quadratic slope for age-squared; b
3jk
,
b
4jk
, and b
5jk
are fixed dynamic effects for parental death, year of
measurement, and young child in the household, respectively; and e
ijk
is the
error term.
At Level 2, I estimated the effects of fixed explanatory variables on the
three random effects (level, and linear and quadratic rates of change) in filial
responsibility across persons within each family cluster. Random effects are
predicted by the following equations:
a
jk
= g
00k
+ g
01k
x
jk
+ r
0jk
b
1jk
= g
10k
+ g
11k
x
jk
+ r
1jk
b
2jk
= g
20k
+ g
21k
x
jk
+ r
2jk
,
where x
jk
describes a fixed, person-specific variable (with implied extension to
the multivariate case) for the jth person in the kth family; g
01k,
g
11k
, and g
21k
are
the within-family estimates of the relationship between
x
jk
and the three random
effects; g
00k
, g
10k
, and g
20k
represent the Level 2 intercepts in each equation;
and r
0jk
, r
1jk
r
2jk
, are error terms.
The Level 1 fixed effects for parental death, year of measurement, and
child in the household have no predictors as they were estimated without
variance under the assumption that their effects do not vary within the same
family:
67
b
3jk =
g
30k
b
4jk =
g
40k
b
5jk =
g
50k
Finally, at the third level of aggregation, the family-level effects are
averaged across the k families. To identify the model, only the three age-
related effects were estimated with variance across families:
g
00k
= z
000
+ u
00k
g
10k
= z
100
+ u
10k
g
20k
= z
200
+ u
20k
,
with the z estimates representing averages of the Level 2 random effects, and u
representing error terms or family variation in the random effects. Estimates
reported in the tables are the z coefficients shown above. In order not to
oversaturate the three-level model, I made the simplifying assumption that the
effects of Level 2 estimates outside of the three age-related intercepts shown
above are considered to be constant across families. Error variances
associated with age estimates (level, linear, and quadratic) at Level 3, together
with corresponding Level 2 error variances, allow computing intraclass
correlations that identify the degree of intrafamily resemblance in levels and
rates of change in filial responsibility.
68
Applying the life course approach presents empirical challenges when
the goal is disaggregating the mutually confounded effects of age, period, and
generation. Solutions to this problem generally involve making simplifying
assumptions about one or more of the effects in the estimated model. I used
several simplifying strategies based on theory and previous literature to ensure
identification of our model. These included the decision to treat the period
effect as a constant across individuals, the use of adjacent generations that
have some degree of age overlap to represent generations, and the creation of
a synthetic lifespan model using individuals from different stages of life (for
other applications of these strategies see, Horn & McArdle, 1980; McArdle et
al., 2004; Roberts & Bengtson, 1999).
D. Results
Summary statistics for age and filial norms over time are shown in Table
3.1. I note that norms appear to be weakening with increasing age of the
sample. Because these statistics are based on aggregations, however, it is not
possible to attribute this trend to intrasubject development, or aging. I next
present the estimates from the hierarchical linear model in Table 3.3. The
model is shown in three progressive steps. First, the unconditional model is
shown; that is, a model with only random age effects at Level 1 and no effects
at Level 2. Second, the unconditional model is shown with two fixed effects
added to the random age effects at Level 1. Third, the conditional model is
shown, adding Level 2 variables to predict variation in the random age effects.
69
Table 3.3.
Maximum Likelihood Estimates for Three-Level Growth Curve Model Predicting Filial Norms
Across 15 Years within Family Context (N = 1,627)
Filial Norms
Model 1 Model 2 Model 3
Fixed Effects Unstandardized
Coefficient
SE Unstandardized
Coefficient
SE Unstandardized
Coefficient
SE
Intraindividual
Level
Status at average
age
14.96*** .12 14.87*** .12 14.86*** .16
Linear rate of
change
-1.11*** .06 -.64*** .08 -.13 .26
Quadratic rate of
change
-.37*** .04 -.32*** .04 -.36** .10
Death of parentsª -1.63*** .18 -1.44*** .22
Decade of
measurementª
-.43** .13 -.99*** .26
Child in
householdª
.10 .16 .10 .15
Interindividual
Level
Status at average
age
70
Table 3.3, Continued
Female
.50* .19
College educated
-.49* .22
G3 cohort
1.54** .46
Support to parents .91*** .21
Divorced -.08 .22
Parents divorced -.31 .26
Parents marital
unknown
-.18 .32
Linear Rate of
Change
Female -.32** .12
College educated -.02 .12
G3 cohort -.24 .48
Support to parents .12 .14
Divorced .16 .13
Parents divorced .29 .16
Parents’ marital
unknown
-.26 .17
Acceleration of
Change
Female
-.02 .06
College educated
-.03 .07
G3 cohort
.07 .20
71
Table 3.3, Continued
Support to parents
.01 .07
Divorced
.08 .19
Parents divorced .06 .08
Parents’ marital
unknown
-.07 .10
Note: Predictor variables are mean-centered. Effects are estimated with robust standard errors.
ªDenotes fixed effect estimate with interindividual variance restricted to zero.
*p < .05. **p < .01. ***p < .001.
I note that all estimates account for family clustering (Level 3) as noted by the
family-level variance reported in the table.
The first equation shows estimates for the unconditional model. The
predicted level of filial responsibility for the average-aged respondent was about
15 points on the 0 - 24 scale. Slope estimates associated with age and age-
squared were negative, implying an accelerating downward trend in the
strength of filial responsibility with increasing age. Using these coefficients, I
calculated the inflection point of this curve--the age at which the slope starts to
move in a negative direction—to be 37.5 years. Thus, I see no evidence of a
midlife peak in filial norms, and, indeed, find that these norms begin weakening
at the start of middle age with the decline accelerating through old age.
72
The second equation shows the unconditional model with Level 1 fixed
effects for parental death (both parents deceased), period of measurement
(1985 vs. 1994, 1997, 2000), and the presence of young children. Both slope
estimates associated with age were negative, again implying an accelerating
downward trend in the strength of filial responsibility with increasing age. The
inflection point shows that norms peak at 42.3 years, after which the slope
reverses to a negative direction. The fixed effect related to parental death
enabled us to compare within-subject change from the time that at least one
parent was alive to the time that both parents were deceased. The estimate
shows that following generational succession, the strength of filial norms
significantly weakened by an average of 1.6 points, a finding consistent with the
notion that filial norms are reevaluated when the possibility of being a care
recipient becomes more real. This effect is specific to each individual as it
depends on the age that individual was when experiencing the death of the
second person and thus going through the process of generation succession.
The fixed effect for period of measurement suggests that filial norms weakened
over historical time, significantly declining between 1985 and the decade of the
1990s. I did not find that the ebb and flow of role conflict, in the form of children
under the age of 18 entering and exiting the household, produced a significant
change in filial norms.
The third equation presents the conditional model with the addition of
Level 2 predictors. In this model, the age coefficients once again revealed an
73
accelerating rate of decline in the strength of filial norms over the lifespan, with
the decline beginning at 51 years. Our findings showed that gender,
generation, a history of providing care to parents, and education level had
significant effects on growth parameters. As illustrated in Figure 3.1, at the
Figure 3.1
Growth curves for filial norms by age predicted for male and female (N = 1,627)
0
2
4
6
8
10
12
14
16
18
20
22
24
18 23 28 33 38 43 48 53 58 63 68 73 78 83 88
Age in Years
Filial Norms
Male Female
intercept age of 52.5, women averaged about half a point higher than men on
the filial norms scale (remaining higher than men at every age), and had a
74
significantly more negative instantaneous slope than men. Taken together,
these findings revealed that filial norms of women--though consistently higher
than men and similarly nonlinear--peaked in strength earlier in life before
declining more rapidly.
The two generations differed significantly in their levels of filial norms at
the intercept, when each reached the comparable age of 52.5 years. As
illustrated in Figure 3.2, those from the later-born generation (G3) scored about
Figure 3.2
Growth curves for filial norms by age predicted for earlier-born generation (g2) vs. later-born
generation (g3) (N = 1,627)
0
2
4
6
8
10
12
14
16
18
20
22
24
18 23 28 33 38 43 48 53 58 63 68 73 78 83 88
Age
Filial Norms
Generation 3 (later born)
Generation 2 (earlier born)
75
one and a half points higher on filial norms than did those from the earlier-born
generation (G2). I note from preliminary analyses that the generational
difference emerged only when historical period was included in the model
suggesting that historical period suppressed the ability to detect the familistic
tendencies of the G3 generation (who reached their early fifties in the less
familistic 1990s).
I found that individuals with no college experience had significantly
higher levels of filial norms than those with at least some college education. As
illustrated in Figure 3.3, individuals with college education scored about half a
point lower on their levels of filial norms at midlife when compared to those with
less education. I further found that those who had the experience of providing
household, transportation, or shopping support to a parent had significantly
stronger filial norms at midlife when compared to those who never provided
such support. This effect is shown in Figure 3.4. Somewhat surprisingly,
marital disruption of the adult child and the parent had little effect on the level
and growth rate of filial norms.
Finally, I addressed the question of how common family membership
influenced filial norms. I addressed this issue first by decomposing the variance
of the raw data, and then the variance of the random parameter estimates in the
growth model. When partitioning the total variance in filial norms across the
three levels of analysis, we found that 48% of the variation occurred within
individuals (over time), 47% occurred between individuals (within families), and
76
Figure 3.3
Growth curves for filial norms by education level (N = 1,627)
0
2
4
6
8
10
12
14
16
18
20
22
24
18 23 28 33 38 43 48 53 58 63 68 73 78 83 88
Age
Filial Norms
College Educated
Less than College Education
only 5% occurred between family aggregates. These results suggested that
family members are quite heterogeneous in their expression of filial norms, and
that family units are not very discernable in this regard. Families are more
internally diverse than they are different from each other.
77
Figure 3.4
Growth curves for filial norms by caregiving experience (N = 1,627)
0
2
4
6
8
10
12
14
16
18
20
22
24
18 23 28 33 38 43 48 53 58 63 68 73 78 83 88
Age
Filial Norms
No Prior Caregiving
Prior Caregiving Experience
The variance decompositions for the growth parameters are shown in
Table 3.4. For each random parameter, I calculated the intraclass correlation, a
statistic that expresses the percentage of variance in the parameter estimates
that can be attributed to between-group variation. Level 1 error variance is not
considered in this calculation because the focus is on the random parameter
and not the data. Higher intraclass correlations demonstrate greater within-
78
Table 3.4
Variance Decomposition from the Three-Level Analysis of Filial Norms
(N = 1,627)
Variance of Estimates
Variance Components
Model:
Unconditional
Model
Model 2:
Unconditional
Model with Fixed
Effects
Model 3:
Conditional
Model with
Fixed Effects
Level 3
Intercept 1.09*** 1.24*** 1.22***
Linear Slope .14** .10** .10*
Quadratic Slope .04** .04** .05**
Level 2
Intercept 7.58*** 7.34*** 6.89***
Linear Slope 1.26*** 1.22*** 1.04***
Quadratic Slope .09*** .09*** .08***
Level 1 (no tests of
significance)
8.07 7.93 7.92
79
Table 3.4, Continued
Intraclass correlations (no tests of significance)
Intercept .13 .14 .15
Linear Slope .10 .07 .09
Quadratic Slope .30 .30 .38
*p < .05. **p < .01. ***p < .001.
family resemblance. Across the three models, we found that 13% - 15% of the
variance in the random intercepts, 7% - 10% of the variance in the linear
estimates, and 30% - 38% of the variance in the quadratic estimates result from
between-family differences. These correlations generally increase with the
complexity of the model as other sources of variation are controlled. In the final
model, almost a tenth of the variance in the linear estimates and one third of the
variance in the quadratic estimates (Model 3) occurred between family units,
providing evidence that lifespan patterns of change in filial norms are to some
degree shaped by common membership in distinctive family contexts.
Nevertheless, the overall results suggest that family groups are internally
diverse with respect to norms, as most of the variation is found between
members in the same families.
80
E. Discussion
I began this investigation by asking the following general question: How
do filial norms for supporting older parents change over the life course? Guided
by the principles of the life course perspective, the paper focused on dynamics
related to the process of aging, the passage of historical time, and generational
succession both in families and in society. Several of the only theoretical
guideposts in this area of inquiry specify that normative obligations toward older
adults peak in midlife, either as a manifestation of filial maturity or as a reaction
to parent-care anxiety. Empirical examinations (e.g., Guberman, 2003; Rossi &
Rossi, 1990), however, almost all using cross-sectional data, have generally not
supported such a prediction, and in fact, showed the very opposite. Taking
advantage of a multiwave, multigenerational data set, I was able to gain some
leverage over the challenges posed by traditional data limitations to examine
competing social forces that shape norms of familism over the life course. To
elucidate our findings with respect to the effects of aging, I present in Figure 3.4
a summary of the predicted age trajectories from our three estimated models.
The unconditional model shows little to support the thesis of midlife
exceptionalism, mainly showing an accelerating decline in filial norms with age.
The shapes of the next two curves, however, reveal a progressive moderating
tendency that is mirrored in the age at which the curves peak. The inflection
point of the curve rises with the complexity of the models, going from 38 in the
81
Figure 3.5
Growth curves for filial norms by age predicted by three models (N = 1,627)
0
2
4
6
8
10
12
14
16
18
20
22
24
18 23 28 33 38 43 48 53 58 63 68 73 78 83 88
Age
Filial Norms
Model 1 Model 2 Model 3
Note: Model 1 is completely unconditional; Model 2 is unconditional with fixed Level 1 effects;
Model 3 is conditional on Level 2 covariates with fixed Level 1 effects (see Table 3). Vertical
bars trim the outlying 2% of the age range in the sample.
unconditional model, to 42 with the addition of fixed effects, and then to 51 in
the conditional model. Once social forces associated with age are controlled,
82
the average age at which norms peak is squarely in middle age, approximating
the lifespan metabolism that developmental theories predict.
The norms of young adults are relatively strong, especially when
contrasted with sharp declines between midlife and old age. Guberman (2003)
explains similar findings by noting that the young are typically far removed from
any need to care for older relatives, and thus present a more idealistic view of
caregiving without considering the practical implications of what such
responsibility may entail. Older people may already need help or see
themselves as needing help in the near future and better appreciate the
sacrifices involved. It is quite reasonable that the shift in self-perception from
potential provider to potential receiver of care promotes the altruistic goal of
saving children from the burden of care. Buttressing this interpretation is our
finding that generational succession weakens norms of filial responsibility. In
results not shown, I found that the death of one parent does not induce a
change in filial norms, affirming that it is only full succession into the oldest
generation (i.e., the death of both parents) and the accompanying change in
perspective that diminishes the norm of filial duty to older parents. As
suggested by Rossi and Rossi (1990), normative obligations to older parents
may also tend to weaken among those who have no relevant kin. The effect of
generational succession is individual to each subject as it depends on the age
in which that individual experienced the death of the second parent, thus
completing the process of generational succession. In order to illustrate the
83
effect of generational succession, I chose to discuss a hypothetical average
subject. Let’s assume the subject was 60 when he lost his second parent and
thus became the oldest generation in his family. His individual growth curve, as
shown in Figure 3.6, will show a peak at age 51, followed by a slight decline
and a drop at the age of 60 as the decline in his normative expectations
accelerated due to his generational shift in the family lineage.
Figure 3.6
Growth curve for a hypothetical average male who is going through generational transition at
the age of 60
0
2
4
6
8
10
12
14
16
18
20
22
24
18 23 28 33 38 43 48 53 58 63 68 73 78 83 88
Age
Filial Norms
84
Findings with regard to historical change in filial norms revealed that filial
norms weakened from the 1980s to the 1990s, giving credence to the claim
made by some scholars (e.g., Bellah et al., 1985; Putnam, 1995) that the family
as a social institution is in decline. This decline is mirrored in reductions over
the same historical period in the valuation of other family relations such as
marriage (e.g., Thornton & Young-DeMarco, 2001), in global values of
collectivism (Roberts & Bengtson 1999), and, more specifically to old age, in the
proportion of older persons living with their families (Casper & Bianchi, 2002).
Contrary to expectations, the later-born generation was more familistic
than the earlier-born generation. In other words, the generation born in the
1950s and 1960s had stronger filial norms in midlife than their parents a
generation earlier. The profamilistic trend across successive generations exists
simultaneously with an opposite historical trend of weakening norms; this
implies that when the baby-boom generation reached midlife in the 1990s, they
were more familistic than the period they aged into. These opposing trends are
illustrated in Figure 3.7. The figure describes two hypothetical individuals. The
first is an average male from the earlier-born cohort (G2) who, as the average
member of his generation in this study, turned 62 in 1990, and thus experienced
a drop in levels of filial norms when entering the less familistic 90s. The second
is an average male from the later-born generation (G3) who, as the average
member of his generation in this study, turned 40 in 1990, and thus experienced
a drop in levels of filial norms when entering the less familistic 90s. Both
85
members of both generations experienced a drop when entering the less
familstic 90s. However, because the member of the later-born generation (G3)
is more familistic than his father from the earlier-born generation (G2), he would
still report higher norms at any age when compared to his parent from the
earlier-born generation. A similar cross-cohort trend was found by Alwin (1998)
with regard to attitudes toward coresidence with aged parents. One way to
search for explanations is to ask whether there were any significant changes in
social policy in the 80s or 90s that may have affected the two generations in a
differential way. For example, it is conceivable that the surge in institutional
forms of long-term care for older adults (peaking in the 1970s and 1980s)
represented a legitimation of formal service use in the middle-aged G2
generation that to some degree dissipated in the G3 generation by the time they
reached middle age in the 1990s. Other researchers have attempted to explain
such opposing findings in a different way. Using the same data set as this
current investigation—the Longitudinal Study of Generations, Bengtson et al.
(2002) used a cohort-sequential design to compare baby boomers and
generation x in a wide variety of characteristics including transmission of
values. Their results revealed mixed trends showing, on the one hand,
evidence to family decline, and on the other, evidence to family resilience and
strong solidarity. They termed these opposing trends “the paradox of continuity
and change across generations” (p. 166). Bengtson et al. (2002) concluded
that the family represents a balancing force that allows for the co-existence of
86
continuity and change in the human society over time. Based on their view, the
conflicting results in this study may be interpreted to suggest that families
represent a force that protects its family members from historical change.
Families are still able to produce familistic children even in less familistic eras.
Figure 3.7
Growth curves of two hypothetical average male subjects; The first, a male from generation 2
(earlier born) who started the 90s at age 62. The second, a male from generation 3 (later-born)
who started the 90s at age 40.
0
2
4
6
8
10
12
14
16
18
20
22
24
18 23 28 33 38 43 48 53 58 63 68 73 78 83 88
Age
Filial Norm
Generation 2 (earlier-born) turn 62 in the 90's
Generation 3 (later-born) turn 40 in the 90's
As expected, I found that women consistently express stronger filial
norms than men from young adulthood to old age. The strength of norms,
however, begins to weaken earlier in life for women than for men. If women are
87
socialized to find greater value in caring roles, then their greater enactment of
this role with respect to parents and other relatives may induce in them a more
realistic appreciation of the sacrifices required of such efforts, and the demands
it places on adult children.
Despite findings in the literature that family disruption and remarriage
have significant effects on generalized filial norms, personal expectations
regarding parental care, as well as actual support behaviors (e.g., Ganong &
Coleman, 1999), I found no evidence that divorce in either generation had any
bearing on normative prescriptions about parent care. A more nuanced
approach that takes into account timing of divorce and remarriage—unavailable
in our data—may provide a better model for detecting such effects in the future.
The concept of filial responsibility as the normative aspect of filial support
says little about the expectations that adult children have with respect to their
own parents. Such personal expectations are likely to be the mediating link
between norms and caregiving behaviors. In the current study, I did not find
that role conflict, in the form of caring for young children in the household, had
an effect on the temporal rhythm of filial norms toward older adults. It is quite
possible, however, that responsibility to children alters personal expectations
with regard to care for parents, and even more so, inhibits the amount of actual
help given. I found that providing support to a parent was positively associated
with filial norms. Although I cannot authoritatively establish the causal
relationship between norms and the provision of care, it is reasonable to
88
assume that support is a manifestation of held norms as well as their
progenitor.
Relatively low within-family consistency in the level of normative beliefs
about intergenerational support may seem somewhat puzzling given that the
norms I consider are directly related to family outcomes and that families are
typically the main agents of child socialization with regard to value orientations.
Contemporary perspectives on modern family forms tend toward the notion that
family context is just one of several social environments that shape normative
values. Thus, it should not be surprising that siblings, spouses, and
parents/children in the same families remain, to a large degree, independent of
each other in their eldercare norms. Put another way, the family aggregates
are more similar to each other than the individuals within them, possibly
reflecting the ethnic homogeneity of the sample.
Despite relying on normative theories of human development, I consider
the concept of filial maturity to be an overly prescriptive and ethnocentric
concept. In societies characterized by chronic poverty, adult children may
choose to devote scarce resources to their children rather than to aging parents
(Aboderin, 2004). It would be loath to label such behavior as immature. It is
perhaps in the United States, where policies toward older people are fairly
restrictive but where the society at-large is fairly wealthy, that norms for
eldercare are most apt to peak in midlife. Approaches that compare family
89
ideals and practices across diverse nations and cultures will be needed to
ultimately test the universality of the midlife maturity model.
Although the sample used in this investigation provides several
advantages for the research questions posed, in terms of its longitudinal design
and inclusion of multiple family members, it is notably absent of minority
families, clearly a limitation in the study of filial norms. Still, one may speculate
that Black, Hispanic, and Asian families would more strongly subscribe to filial
norms, given cultural values that tend to stress kinship obligation. I can only
imagine if minority groups would have had a more gradual decline in norms
over time or introduced greater intrafamily resemblance (because of cultural
similarity) than I observed in the largely White families in our sample, though
these seem to be reasonable expectations. I also note that the sample is not
nationally representative, and lacks the coverage needed for making conclusive
inferences about the American population. These limitations, however, must be
weighed against the unique multiple-time, multiple-generation design of the
study that makes it possible to disaggregate social change into its various
temporal and contextual components. Finally, as in all longitudinal studies,
missing data in repeated measures are cause for concern regarding biases
introduced by systematic attrition and patterning of incompleteness. Although
such biases can never be formally ruled out, I found little evidence that attrition
or patterned incompleteness affected the growth-curve trajectories modeled in
our analyses.
90
In conclusion, the results show that filial norms toward care of older
adults demonstrate considerable malleability across the adult life span, a
pattern that is most consistent with the lifelong openness model. Indeed, the
change is occurring in late life when reappraisals are likely the result of altruism
(growing relevance as a potential receiver) or role loss (growing irrelevance as
a provider). Eldercare norms appear to be sensitive to exigencies of personal
circumstances, emphasizing the notion that norms toward the care of older
people are linked to situational factors, and are to some degree reciprocally
related to them. Would more abstract, universalistic norms about the salience
of family life be more trait-like and, thus, less sensitive to objective conditions
than norms that are tagged to a particular target person? I hope that our
research opens the door to addressing such a question by providing an
inclusive framework for understanding the multidimensional and dynamic nature
of filial norms.
91
Chapter Four: How Parents Influence Their Children to Provide Support
to Them In Old Age?
A. Introduction
The need for prolonged periods of family caregiving to older adults has
grown with declines in mortality rates over the past century (Siegel, 1993), and
will continue to grow with increasing life expectancy at the end of the life span
(Crimmins, 2004). Parental caregiving has become a normative part of the life
course of middle aged persons, and particularly women (Sorenson & Zarit,
1996; Zarit & Reid, 1994). While previous research has documented the
willingness of adult children to care for aging family members (Brody & Brody,
1989; Cicirelli, 1988, Stone et al., 1987), only a few studies (i.e., Silverstein et
al., 2006; Silverstein et al., 1995) investigated the motivations and normative
obligations that guide such supportive behaviors. The purpose of this
investigation was to examine whether parents can influence the amount of
support they will be receiving in old age from their adult children by affecting
their children’s motivations, normative obligations, or actual supportive
behaviors throughout the life course. Specifically, I examined four direct
pathways through which parents may influence children’s supportive behaviors:
(1) socialization of normative behavior; (2) social learning through modeling; (3)
solidarity and strong emotional bonding; and (4) intergenerational social
exchange across the life course in two possible forms: providing support in the
form of in-vivo transfers to children earlier in life as a form of “prepaying” for
92
support in old age; or promising bequest to children as a form of “repayment”
for parental care in old age. In addition, I investigated the possible role of
frequency of contact between children and mothers and geographical proximity
between children and their mothers and their role as mediating factors,
connecting each of these four pathways indirectly to actual supportive
behaviors.
Theoretical and Empirical Explanations for Intergenerational Resemblance in
Normative Orientations and Behaviors
In sociological literature, socialization or social learning theories with the
concept of role-modeling, as well as exchange theories and the solidarity
model, offer alternative explanations of intergenerational resemblance in
normative orientations and behaviors. This section presents a brief overview of
each of these models.
Socialization
Traditional view of socialization suggests that the family, and in particular
parents, are the principal agent of socialization through which children are
taught the norms of social order. Attitude similarity across generations within
the same family would thus represent a successful parental socialization of
norms and beliefs (Glass, Bengtson & Dunham, 1986; Miller & Glass, 1989).
However, previous studies found weak effects of parents-children transmission
of social norms (e.g., Bengtson, 1975; Glass et al., 1986; Hoge, Petrillo &
93
Smith, 1982) as well as weak correlations between parents’ and their children’s
values (e.g., Bengtson, 1975; Whitbeck & Gecas, 1988). In the context of this
investigation, attitude similarity in normative obligations towards parental care
between adult children, as currently stated, and their parents, as stated earlier
in time, may represent successful socialization of such norms.
Social learning theories
Social learning theories specifically emphasize the role of modeling in
socializing children to various desired behaviors. In the context of filial norms,
parents’ support to their own parents will provide a model to their children, and
their children’s willingness to provide such parental support will demonstrate the
degree to which they are willing to imitate that model (Ribar & Wilhelm, 2006).
Results from previous studies on the role of behavior modeling are mixed.
Several empirical studies revealed evidence in support of social learning
theories. For example, using nationally representative data from the National
survey of Families and Household (NSFH), Szinovacz (1997) found that a
respondent’s attitudes about letting aging parents live with him/her was
positively related to whether grandparents lived with his/her parents during
his/her childhood. Using the three generations study of Mexican Americans,
Ribar and Wilhelm (2006) found that the attitudes towards co-residence with
elderly parents and towards the financial responsibility to aging parents of the
youngest generation were affected by help the middle generation provided to
the oldest generation. However, using a sample of women from the National
94
Longitudinal Surveys, Caputo (2005) suggested that parents’ caregiving
behavior to their own parents did not influence their daughters’ caregiving
activity.
Exchange theories and intergenerational reciprocity
This paper suggests to distinguish between two possible ways in which
intergenerational exchange and reciprocity may take place. The first is based
on sociological theories of exchange and on the notion that reciprocity is
enforced by a normative obligation to repay for a previous debt (Silverstein,
2005). In other words, parents’ transfers to their children earlier in life will be
reciprocated through the children’s inclination to provide support to them later in
life (Silverstein, Conroy, Wang, Giarrusso & Bengtson, 2002). Silverstein et al.
(2002) discussed two different models that follow this line of thought: an
investment model—a return on investment made earlier in life, and an
insurance policy—earlier transfers to the child are reciprocated by the child
when the parent encounters specific conditions in old age; as well as other non
reciprocal models such as altruism. The second is based on economic theories
of rational choice (e.g., Coleman, 1990), suggesting that people‘s behavior is
guided by self-interest and is based on analysis of costs and benefits. Caputo
(2005), for example, used this theoretical framework to explain his finding that
inheritance may be a motivation for an adult daughter’s decision to provide
support to her parents.
95
Several theoretical concepts have been used to describe the concept of
reciprocity in intergenerational transfers across the life span. Silverstein, et al.
(2006) discussed the concept of filial norms as an aspect of social capital that
resides in family relationships. They discussed social capital as a latent
resource that accrues when others feel the obligation to reciprocate and provide
something in return for earlier investments. Social capital was defined by
Furstenberg and Kaplan (2004) as the “stock of social goodwill created through
shared norms and sense of common membership” (Furstenberg & Kaplan,
2004, in Scott, Treas, & Richards, p. 221). An earlier model--the support bank
model (Antonucci, 1990) suggested a similar notion. Support bank is a reserve
that parents build early in life through investing money, time, and other
resources in their children. Later on in life, parents may draw from this support
bank by receiving support from their adult children in times of need (Silverstein,
2005; Silverstein et al., 2006). Yet another similar concept to discuss
intergenerational reciprocity as a form of social exchange across generations is
the concept of latent kin matrix. Coined by Riley (1983), this term represents
the existence of family ties that may remain dormant for long periods of time but
may become active as a source of support in times of need. Bengtson (2001)
utilized this concept in his explanation of the increasing role of extended family
members in the lives of modern families and especially the potential role of
grandparents in raising grandchildren. In the context of this paper, this concept
96
may be applied to viewing adult children as a latent kin matrix, providing
support to their aging parents only when the needs arise.
Affectual solidarity, normative solidarity, associational solidarity, and supportive
behavior
Several studies have investigated the association between filial norms
and the actual provision of support. Such association is suggested by a larger
theoretical model – the intergenerational solidarity model. Intergenerational
solidarity (Bengtson, 1996; Parrott & Bengtson, 1999) is a multidimensional
approach representing dimensions along which family cohesion or integration
occurs (Parrott & Bengtson, 1999). These dimensions include the following
constructs of solidarity: affectual--the positive sentiment toward other members
of the family; associational--the frequency and patterns of interaction;
consensual--the degree of similarity in values, attitudes, and opinions;
normative—the strength of commitment to perform familial roles and to meet
familial obligations; and functional—the degree to which family members
actually exchange support (Bengtson & Mangen, 1988; Lee et al., 1994).
Several studies (e.g., Bengtson & Roberts, 1991; Parrott & Bengtson, 1999;
Whitbeck et al., 1994) have examined interrelations among subsets of the
intergenerational solidarity dimensions and the empirical evidence confirms the
theory and demonstrates its utility in examining the nature and the quality of
relationships between parents and adult children (Allen et al., 2000).
97
Normative obligation and actual supportive behavior: Several studies focused
on the specific association between normative and functional solidarity (Lee et
al., 1994). Bromely & Blieszner (1997) suggested that the sense of filial
responsibility an adult child feels to provide support for his/her aging parent
affects the actual caring for that parent. Silverstein & Litwak (1993) provided
empirical evidence which suggests that stronger filial obligation increases the
likelihood that adult children will provide instrumental support. Furthermore,
Silverstein et al. (2005) found that among sons, stronger endorsements of
norms of familism are associated with greater support. Peek et al., (1998)
suggested that parents might modify their social norms regarding receiving care
from adult children (global filial responsibility expectations from adult children)
based on their children’s circumstances and resources. Using data on impaired
elderly parents who lived in the community, the authors found such adjusted
norms (termed specific expectations or behavioral intentions) influence the
amount of care received from the adult children. Whitbeck et al. (1994) found
that filial concern influenced intergenerational help and support exchanges.
Willingness to help and support a parent was, in part, mediated by concern for
the parent and a sense of duty. In summary, normative and functional solidarity
were found to be associated. More specifically, filial norms, held by adult
children as well as filial expectations held by aging parents were found to be
associated with the amount of help and support provided to aging parents by
adult children in times of need.
98
Emotional closeness and actual supportive behavior: Other studies focused on
the aspect of affectual solidarity—the sense of closeness to parents, and its
effects on functional solidarity—the actual supportive behaviors. For example,
using attachment theory as a framework for studying helping behavior towards
parents, Cicirelli (1983) found that emotional closeness to parents affects future
helping behavior directly. Rossi and Rossi (1990) found similar results
suggesting that affective ties affect exchange behaviors in both directions.
Support to the significant role emotional closeness plays in supportive behavior
was further provided by Silverstein et al. (2002), who found that emotional
closeness triggers higher levels of support and suggested that fostering close
emotional bonds may be used by parents as an investment that may lead to
increased future caregiving.
Contact with parent and actual supportive behavior: Associational solidarity, or
the amount of contact between adult children and their parents, may play an
important role in predicting future support. Cicirelli (1983) referred to
communication with parents via phone, mail, visits etc. as attachment behaviors
that allow adult children to feel a sense of psychological closeness even if they
are no longer physically close to them. His study suggested that the frequency
of such communication between adult children and their parents plays a
significant role in mediating both the effects of emotional closeness and
normative obligations on actual helping behaviors. In fact, Cicirelli (1983)
concludes that inducing contact between adult children and parents would
99
serve as an “indirect means of stimulating help to parents” (p.823). This study
puts to the test the role of associational solidarity as a possible mediating factor
between all four pathways of possible influences studied in this investigation:
socialization of normative solidarity, role modeling of supportive behaviors,
affectual solidarity, and exchange and reciprocity.
This study is designed to investigate these mechanisms as they
contribute simultaneously to both instrumental types of support (i.e., help with
household chores, transportation etc.), and emotional-social types of support
(providing companionship, information and advise etc.). Distinguishing between
these two types of support is important especially when attempting to
understand what motivates adult children to provide such support. It is likely
that the mechanisms that lead to each of these support types are different.
Recognizing that gender is an important factor in shaping the relationship
between children and parents, this analysis focuses on mothers only and
controls for gender of the adult child in every analysis.
B. Research Questions and Expectations
This investigation addresses the basic question of what motivates adult
children to provide support for their aging parents. It specifically asks whether
there are mechanisms through which parents can influence the amount of
support they will be receiving in old age from their adult children by affecting
their children’s motivations, normative obligations, or actual supportive
100
behaviors throughout the life course. Based on the literature review and the
theoretical examination, I present a conceptual framework (see Figure 4.1) that
suggest four alternative, yet possibly complementing, pathways.
The paper asks questions and makes predictions with regards to four
such possible mechanisms both in direct and indirect ways as mediated by
frequency of intergenerational contact and intergenerational geographical
proximity:
Socialization: The paper asks whether parents can socialize their children to
provide support to them by instilling high levels of elder care norms in them. If
indeed parents, as the primary source of socialization, can affect the
development of eldercare norms in their children, we expect to find similarity in
normative orientations towards elder care between adult children, as expressed
in the present, and their parents, as measured in earlier period of time, when
the parents were of similar age to that of the adult children in the present. I
further expect to find that such filial norms will, in turn, lead to higher levels of
instrumental and/or emotional support. In light of previous findings (e.g.,
Cicirelli, 1983) that emphasized the role of communication and contact between
adult children and parents in mediating the relationship between normative and
helping behaviors, I expect to find a direct relationship between norms and
101
Figure 4.1
Conceptual and Empirical Causal Model
Support
to
mother
(Instrumental
and Emotional)
2000
Emotional
closeness to
mother
1985
Expectation for
inheritance
1997-2000
Received financial
support from
parents
1971-1994
Mother demonstrated
support to older
relatives
1971-1985
Emotional
closeness to
mother
2000
2. Social learning
3. Affectual Solidarity
4. Exchange and
Reciprocity (a) prepayment
(b) repayment
Mother’s filial
Norms
1985
Child’s filial Norms
2000
1. Socialization
Contact
with Mother
& Proximity
support as well as an indirect effect mediated by frequency of contact with
parents and/or close geographical proximity to them.
Social learning through modeling: The paper asks whether parents can affect
their children’s level of support to them in old age by modeling to them such
care through providing care to their own parents when the children are younger.
I hypothesized that adult children whose parents provided support to their own
102
parents (the children’s grandparents) would exhibit higher levels of parental
supportive behavior in the present and that the link between witnessing such
support in the past and present levels of parental support would be direct and
indirect through higher levels of filial norms. In other words, I expect that those
adult children would develop higher levels of filial norms than those who did not
witness such parental care. Higher levels of filial norms would, in turn, increase
parental support either directly or indirectly through frequency of contact. In
addition, and in light of previous findings (e.g., Cicirelli, 1983) that emphasized
the role of communication and contact between adult children and parents, I
investigate whether witnessing such intergenerational support leads to closer
patterns of communication and to living closer to one’s parent, which, in turn, is
expected to lead to higher levels of support.
Affectual solidarity and emotional bonding: I would expect that adult children
who have stronger emotional bonds with their mothers will provide more
support to them. I would further expect that strong affective/emotional solidarity
in the past will increase the level of such solidarity in the present, thus leading,
indirectly, to higher levels of supportive behaviors to mothers. In addition, as
suggested by previous findings (e.g., Cicirelli, 1983), such emotional bonds or
emotional attachment may manifest itself in attachment behaviors such as living
in close proximity to one’s parent or having frequent communication with a
parent. These attachment behaviors were found to mediate the relationship
between such bonds and caregiving behaviors. I am, therefore, testing for an
103
indirect relationship between affectual solidarity and supportive behaviors as
mediated by contact with and proximity to mothers.
Intergenerational social exchange across the life course: In line with life-course
social exchange theories, I am testing for two possible complementing
scenarios. First, I ask whether children who received higher levels of in-vivo
transfers from their parents in the past would exhibit higher levels of supportive
behavior in the present. Parents, therefore, may affect future care received
through “prepaying” via major gifts or financial support when the children are
younger. Second, I ask whether children who expect to receive bequests
following parental death would exhibit higher levels of supportive behavior in the
present. Such finding may suggest that parents’ promise of a future
“repayment” for present supportive behaviors may be exchanged for current
parental support. I am testing for both direct effects as well as indirect effects
through contact with parents and through living in close proximity to parents. In
other words, I am suggesting that frequency of contact and/or geographic
proximity may play a significant role as a mediating factor here as well, whereby
both mechanisms of “prepayment” and “repayment” may work through inducing
higher frequency of contact or even possibly greater geographic proximity,
which, in turn, will increase higher levels of supportive behaviors.
Associational solidarity: close geographic proximity and frequency of contact:
Associational solidarity is viewed in this analysis as a mediating variable,
hypothetically connecting all four above-described mechanisms with actual
104
supportive behaviors. I hypothesized that more frequent levels of
communication with one’s mother, and living in closer proximity to the mother,
each independently, would lead to higher levels of supportive behaviors.
C. Methods
Sample
A sub-sample of The University of Southern California Longitudinal Study
of Generations (LSOG) will be used for this analysis. The LSOG began in 1971
(wave 1) as a mailed survey with an original sample of 2,044 respondents ages
16 to 91 from 328 three-generation families that were selected via a multi-stage
stratified random sampling procedure from a population of 840,000 individuals
enrolled in southern California's first large HMO. The response rate for wave 1
individual family members was 65 percent, a rate comparable to initial response
rates of other panel surveys of aging. In 1985 (wave 2) effort were made to
locate original sample members, yielding a sample of 1,331, with a response
rate of 62%. Since 1985, data have been collected at three year intervals, with
additional information gathered in 1988 (wave 3) for a total sample of 1,483
(response rate of 66%); 1991 (wave 4) for a total sample of 1,532 (response
rate of 74%); 1994 (wave 5) for a total sample of 1,682; 1997 (wave 6) for a
total sample of 1,724, and 2000 (wave 7) for a total sample of 1,896 (Parrott &
Bengtson, 1999). The LSOG currently has four participating generations - G1
105
(the oldest) through G4 (the youngest) and is preparing for wave 8 of data
collection.
The sub-sample for this analysis is comprised of dyads of G3 children
and their corresponding G2 mothers. The dyads were included if (a) the
mothers participated in the survey at least in 1985, (b) the mothers were alive in
2000, and (c) the adult children participated at least in the 2000 wave. The
selection of the sub-sample based on these selection criteria resulted in 379
adult children and 237 mothers, which in turn resulted in 379 adult child -
mother dyads. The sub-sample includes 131 adult children who are either only
children or single respondents from their family, and 248 respondents who have
at least one sibling in the sub-sample. In terms of gender specific dyads, 222
(58.6%) are daughter-mother, and 157 (41.4%) are son-mother. The average
age of children and parents in 2000 is 47.6 and 71.7, respectively. Table 4.1
provides more information on the characteristics of the sub-sample. I am using
data from the various waves as it applies to the topic at hand and based on
data availability.
Procedure
Analyses were performed using structural equation modeling to estimate
direct and indirect pathways to support provided to mothers. As the first step, a
measurement model was estimated for the latent variables that comprise the
dependent variable. Based on exploratory factor analysis, two latent constructs
106
Table 4.1
Descriptive Statistics: Adult Children (N=379) and Mothers (N=237)
Variable %
Number Mean (SD)
Adult Children (N=379)
Age 47.57 (3.42)
Gender
Female 58.6 222
Male 41.4 157
Education (College + ) 42.5 161
Marital Status (Married) 69.9 265
Mothers (N=237)
Age 71.71 (4.54)
Marital Status (Widowed) 22.8 54
Health
Excellent 3 7
Good 12.7 30
Fair 40.9 97
Poor 43.5 103
Provided care to parent 54 128
107
were created, and the measurement model was assessed. Based on the
analyses and data availability, the following two constructs were created:
instrumental support (including help with household chores, providing
transportation/shopping, and providing help when sick), and social support
(including providing information and advice, providing emotional support, and
discussing life decisions). Models were assessed based on the statistical
significance of estimated path coefficients and goodness-of-fit statistics for the
model as a whole. It is important to note that the model includes
measurements at earlier time waves including 1985, 1991 and 1994 and their
lagged effect on instrumental and social support as measured in 2000, as well
as measurements in 2000 and their casual effects on support. Data on filial
normative obligations, or normative solidarity, were collected in 1985 for the first
time and as a result, I cannot go to earlier points in time to retrieve data on this
variable. For continuity and consistency, I am also looking at affectual solidarity
at the same point in time.
Structural equation modeling allows for testing measurements of latent
constructs and the relationships among these constructs in a system of
equations, thus allowing the investigator to test the hypotheses proposed in this
investigation. In addition, structural equation modeling allows treating full data
in spite of missingness. The analyses use data from more than one
measurement time. As a result, there are missing values resulting from non
responsiveness. Structural equation modeling uses the full information
108
maximum likelihood algorithm (FIML), which estimates the distribution of
incomplete data based on the distribution of complete data (Wothke, 2000),
under the assumption of MAR--“missing at random” (e.g., McArdle et al., 2004).
Prior to estimating the structural equation models, other efforts were made to
address missingness due to nonresponse. Specifically, I used data from
previous measurement times to determine variables that tend to be more stable
in older ages such as education level. In addition, I used single imputation
method of hot deck to replace single missing data based on the full data for
education and income. These attempts to address patterns of missing data are
a part of an effort to use complete data available to maximize the use of all
collected data and avoid dismissing whole cases when only some variables
were not available.
In evaluating structural equation models, two factors are considered.
First, the goodness-of-fit of the model is discussed. While the chi square
statistic represents a common measure of fit, it is very sensitive to the size of
the sample with smaller samples more likely to yield significant chi squares
(Maruyama, 1998). As a result, I chose in this paper to also use alternative and
more appropriate goodness-of-fit indices including the Normed Fit Index (NFI),
and Root Mean Square Error of Approximation (RMSEA). These indices are
independent of sample size and thus represent better methods of evaluating the
results in this analysis. Generally, an acceptable model fit is indicated when
NFI > .90 (Byrne, 2001; Maruyama, 1998). With regards to RMSEA, the
109
following guidelines apply: values less than .08 indicate good fit, with values
under .05 represent a very good fit, values greater than .08 and up to .10
indicate a mediocre fit, and values greater than .10 indicate a poor fit
(Maruyama, 1998). Once the goodness-of-fit of the model is established, the
coefficients of the predicted relationships between the variables and the
correlations are examined.
Measures
Dependent variable:
Parental Support: The exchange of help and support by adult children to their
aging parents in 2000 was measured using a nine-item scale that asks: “Do you
provide any of the following types of help and support to your “STUDY” Mother
/father? (a) household chores, (b) transportation/ shopping, (c) information and
advice, (d) financial assistance, (e) emotional support, (f) discussing important
life decisions, (g) visiting/sharing leisure activities, (h) help when she/he is sick,
and (i) assistance with personal care (e.g. help with bathing, dressing)”. For
each item the respondent had to state the frequency on an eight-point scale
ranging from “not at all” to “daily” (with the following values: not at all, once a
year, several times a year, monthly, several times a month, weekly, several
times a week, daily). Two items were excluded from the analysis -- assistance
with personal care and providing financial support -- because they occurred in
very low frequency in this sample. Specifically, only 2.8% of our sample
provided any level of assistance in personal care and only 7.3 % reported
110
providing any level of financial assistance. The remaining seven items were
factor analyzed, and two main factors were loaded. The item of visiting/sharing
leisure activities did not load on any of the factors and was therefore not
included in the analysis. The two main factors divided the remaining six items
into two types of assistance: (1) items that are more instrumental in nature
including help with household chores, providing transportation/shopping, and
providing help when sick; and (2) items that are more social or emotional in
nature including providing information and advice, providing emotional support,
and discussing life decisions. Based on the factor analysis, I created two latent
variables. The first was termed instrumental support and the second was
termed social support.
Independent variables
Normative Solidarity--filial norms: Norms regarding responsibilities of adult
children to aging parents (filial responsibility) in each wave were measured by a
six-item scale that asks: “Regardless of the sacrifices involved, how much
responsibility should adult children with families of their own have: (a) to provide
companionship or spend time with elderly parents who are in need?; (b) to help
with household chores and repairs and/or to provide transportation for elderly
parents who are in need?; (c) to listen to the problems and concerns of elderly
parents and to provide advice and guidance?; (d) to provide for personal and
health care needs of the elderly parent (for example, bathing, grooming,
medication, etc.)?; (e) to provide financial support and /or assist in financial and
111
legal affairs of elderly parents who are in need?; and (f) to provide housing for
the elderly parents who are in need?. For each item the respondent had to
assign responsibility level on a five point Likert Scale ranging between “none” to
“total” (with the following values: none, minor, moderate, major and total). The
responses for six items were summed to come up with a “total degree of filial
responsibility”, and the scale was floored at 0. The index score ranged from six
to thirty. The index score was then recoded to come up with a baseline
measure of filial responsibility ranging from zero to 24, with zero representing
the lowest degree of filial responsibility and 24 representing the highest degree
of filial responsibility.
Demonstration of support to older relatives: Mothers were considered as
providers of support in or prior to 1985 to their elderly parents or other elderly
relatives if they had responded “yes” to any of the following questions in 1985:
“Are you currently providing any kind of caregiving assistance to a family
member, such as your parents, spouse, or an elderly relative, because of a
chronic disability or health-related problem? or “How about during the past 12
years? Were you involved in providing caregiving assistance to one of your
parents, spouse, or an elderly relative at any time during this period?” Mothers
were also included as providers of support if they had indicated that their
parents (mother or father) depended on them for household chores and
transportation.
112
Affectual Solidarity: Emotional closeness to mother: emotional closeness or
intimacy between the child and the mother was measured by using four
questions that represent the affectual/emotional solidarity scale (see Bengtson
& Mangen, 1988) asking: (1) How close do you feel to your mother?; (2) How
good is your communication with your mother?; (3) How well does your mother
understand you?; and (4) How well do you understand your mother?
Responses to each item score on a sic-point Likert scale. The items were
added to create a summative scale ranging from 0 to 24, with 0 representing
lowest level of emotional closeness and 24 representing the highest level of
emotional closeness.
Past support received from parents: adult children were considered as
recipients of a significant financial support were those who reported in 1985 and
1988 that they relied on their parents for financial assistance in 1985 and/or
1988. In addition, adult children were considered recipients of a significant
support if they had responded “yes” to the following question as measured in
1994: “Sometimes individuals GIVE or RECEIVE special assistance or large
gifts to, or from, family members. These may be in the form of cash, stocks,
bonds, or major items such as large appliances, audio/visual equipment,
furniture, or automobile. Have you EVER RECEIVED a major gift from your
“study” parent?”
Normative expectation of inheritance: normative expectation of inheritance was
measured by a single question asking whether respondents agree or disagree
113
with the statement that parents have a responsibility to leave an inheritance to
their children. The items scored on a 4-point scale ranging from “strongly
disagree” to “strongly agree”. For this analysis the item was dichotomized
where “1” indicates that the adult child agrees or strongly agrees with the
statement and thus expects inheritance from his/her parents.
Associational Solidarity--frequency of contact: frequency of contact was
determined based on responses to four items, each with a six-point Likert scale
responses, asking about the frequency of contact with mother in person, via
phone, mail or e-mail, with responses ranging from “not at all” to “daily or more
often”. The item with the highest frequency of contact was included in this
analysis and represents the total frequency of contact of that adult child with
his/her mother.
Associational Solidarity--geographical proximity: geographic proximity was
determined based on a continuous variable indicating the distance in miles from
the child’s residence to the mother’s residence on a 7-point scale ranging from
“more than 500 miles” to “we live together”.
Control Variables
Several control variables were included in the analyses that are known by
previous studies to influence parental support behaviors. For the parents, these
variables included health, marital status, income and education. For the adult
114
children, these variables included gender, income, education, marital status,
and geographical proximity to mother.
Health was measured using a self-rated health report asking “compared
to people your own age, how would you rate your overall physical health at the
present time”. Responses on a four-point Likert Scale ranged from “poor” to
“excellent”. Marital status was operationalized differently for the two
generations. For mothers, marital status was a dummy variable where “1”
indicates widowed and “0” indicates all others. For adult children, marital status
was a dummy variable where “1” indicates married and “0” indicates all others.
Description of the Conceptual and Empirical Model
The hypothetical causal model is presented in Figure 4.1. Based on
previous research and the proposed hypotheses, I am expecting to find five
alternative direct pathways to parental support, as previously described: (1)
socialization with mother’s filial norms in the past leading to adult child’s filial
norms in the present and leading directly to supportive behavior; (2) social
learning with mother’s past support to her own parents predicting adult child’s
present support to his/her parents; (3) affectual solidarity, where past emotional
closeness between adult child and mother predicts present emotional support
and in turn, supportive behavior; (4) exchange and reciprocity whereby past
receipt of financial support from parents will lead to higher levels of support
115
towards parents; and (5) exchange and reciprocity whereby expectations for
bequest will lead to higher levels of support towards parents.
In addition to the direct pathways, I am also testing four additional
indirect paths where by socialization, social learning, affectual solidarity, and
both forms of exchange (prepayment and repayment) operate through
frequency of contact as a mediating variable. Finally, I am also looking to see
whether mothers’ past caregiving to their own parents predicted higher levels of
filial norms in the adult child, which, in turn, would lead to higher levels of
support. It is important to mention that this investigation is interested in
understanding the relationships between all these pathways and both types of
support—instrumental and emotional—simultaneously.
D. Results
Measurement Model: As previously described in the measurement section, I
used exploratory factor analysis to decide what constructs the various items of
parental support should represent. The two main factors that the factor analysis
yielded included: (1) items that are more instrumental in nature including help
with household chores, providing transportation/shopping, and providing help
when sick; and (2) items that are more social or emotional in nature including
providing information and advice, providing emotional support, and discussing
life decisions. Based on the factor analysis, I created two latent variables. The
first was termed instrumental support and the second was termed social
116
support. As a first step in the analysis, I fitted this measurement model to the
data to determine if the latent constructs are adequately represented by the
observed variables. It is important to note that the two types of support are
likely to be correlated and, therefore, I am testing for correlation between the
latent constructs’ error terms. When testing the model, one should examine
both the model fit as a whole and the coefficients that represent the
relationships between the various variables. I am also testing for correlations in
the error terms between two sets of items: help with transportation and chores,
and providing information and discussing life decisions. These two sets seem
to be hypothetically related, as it is possible that similar factor may contribute to
the provision of each set. In evaluating the goodness-of-fit for this model, there
are several indices that are of common use including the chi square, Normed
Fit Index (NFI), and Root Mean Square Error of Approximation (RMSEA). The
chi square statistic for this model was 12.5 with 6 degrees of freedom and was
not significant, indicating a good fit. The NFI equals .994, and the RMSEA
equals .054, both indicating a good fit. I concluded that this measurement
model is acceptable for the purpose of this investigation. Standardized
measurement coefficients and correlations among the latent constructs in the
best fitting model are shown in Table 4.2 and in Figure 4.2. Based on the
coefficients, it is clear that the items in each latent construct do represent that
construct, as was suggested by the exploratory factor analysis. It is further
117
suggested that the two forms of support (instrumental and social) are correlated
yet still represent separate types of support.
Table 4.2
Unstandardized, Standardized, and Significance Levels for Measurement Model (Standard
Errors in Parentheses); (N = 379)
Parameter Estimate Unstandardized Standardized
Measurement Model Estimates
Instrumental support Æ help with chores
1.00 (NA)na .51
Instrumental support Æ provide
Transportation
1.06 (.12)*** .57
Instrumental support Æ help when sick
2.08 (.28)*** .78
Social support Æ help with information
1.00 (NA)na .73
Social support Æ emotional support
1.39 (.12)*** .91
Social support Æ help with life decisions
1.04 (.06) .77
Covariance of of instrumental
support and emotional support
.61(.10)*** .73
Covariance of Residuals help with
chores and help with transportation
.37 (.07)*** .36
Covariance of Residuals of providing
information and help with life decisions
.46 (.14)** .34
Note: χ
2
(6) = 12.50, p >.05; GFI = .99; RMSEA = .05
*p < .05. **p < .01. ***p < .001.
118
Figure 4.2
Measurement model: Standardized Coefficients
Instrumental
Support
Social
Emotional
Support
d1
d2
e1
e4 e5
e6
e2
e3
Chores
Transportation
Sick
Emotional Information
Life
decisions
.73
.51
.77
.73
.71
.36
.34
.57
.77
Structural Model: As discussed in the measurement model, in a structural
equation modeling, two types of factors are considered when interpreting the
results. First, the model fit as a whole, and second, the various coefficients and
correlations among the variables. In evaluating the goodness-of-fit for this
119
model, I am using two indices including the NFI, and RMSEA. The overall
goodness-of-fit of this model indicates an acceptable fit with NFI equals .98 and
RMSEA equals .060.
In structural equation modeling, the mediating variables, the variables
that are hypothesized to be indirectly linking the independent variables to the
dependent variable, are treated as endogenous variables and as such may be
predicted by other exogenous variables including control variables. Table 4.3
describes the coefficients of the various relationships between various
exogenous variables in the model and control variables and the mediating
variables in the model. Table 4.4 represents the coefficients of the various
models as they pertain to the two dependent variables in the model—
instrumental and social support to mothers. Figure 4.3 provides a graphic
presentation of the main results. The following discussion will present the
results as they relate to the five main pathways hypothesized in this analysis:
Socialization and the role of filial norms: As opposed to this paper’s hypothesis,
mother’s past filial norms did not predict the level of current filial norms on the
part of the children. In addition, the results did not provide evidence to suggest
that past supportive behavior on the part of the parents to their own parents
would lead to higher levels of filial norms on the part of the children.
Furthermore, past filial norms of mothers, which may represent mothers’
expectations of care from their children, did not predict proximity to mother,
frequency of contact, or emotional or social support. However, past filial norms
120
of mothers, which may represent mothers’ expectations of care from their
children, significantly predicted instrumental support. What is surprising,
however, is that this relationship is inverse, suggesting that higher expectations
on the part of the mothers predict lower levels of instrumental support. In other
words, for every one unit increase in the scale of mother’s filial norms in 1985,
there is a one unit decrease in level of instrumental support provided to that
mother in the present. Current filial norms did not significantly directly predict
both types of support, and did not significantly predict any form of associational
solidarity (contact with or proximity to mother).
Social learning: As expected, the analyses show that adult children
whose parents provided support to their own parents exhibited higher levels of
both instrumental and emotional supportive behavior towards their parents.
Mother’s past supportive behavior also predicted the level of frequency of
contact with that mother, which, in turn, further increases the level of emotional
behavior. In other words, mothers who provided support to their own parents in
the past, had adult children who kept more frequent contact with them,
compared to their peers who did not care for their parents. Modeling caregiving
behaviors, however, did not increase the levels of filial obligation in adult
children. Modeling, therefore, leads to imitating the behavior without changing
the normative obligation with regards to such caring behavior.
Affectual Solidarity: As expected, adult children who had stronger
emotional bonds with their mothers provided more support to them. Higher
121
Table 4.3
Unstandardized, Standardized, and Significance Levels for Structural Equation Model for the
Mediating Variables (Standard Errors in Parentheses); (N = 379)
Parameter Estimate Unstandardized Standardized
Endogenous Variable
Filial norms in 2000
Mother’s past filial norms .04 (.05). .04
Mother’s past supportive behavior to own parents -.49 (.36) -.07
Child’s gender -.27 (37) -.04
Child’s income -.03 (.04) -.04
Child’s education -.20 (.14) -.08
Child married (vs. not married) .01 (.42) .00
Mother’s income -.05 (.05) -.04
Mother’s education -.07 (.13) -.03
Mother widowed (vs. not widowed) .52(.43) .06
Mother’s poor health -.61 (.23)** -.14
Affectual Solidarity in 2000
Past affectual solidarity .60(.05)*** .56
Child’s gender -.61(.39) -.07
Child’s income -.03 (.04) -.03
Child’s education -.19 (.14) -.03
122
Table 4.3, Continued
Child married (vs. not married) 1.08 (.45)* .11
Mother’s income -.05 (.05) -.04
Mother’s education -.19 (.13) -.06
Mother widowed (vs. not widowed) .50(.45) .05
Mother’s poor health .10 (.24) .02
Frequency of Contact in 2000
Past mother’s norms .01(.01) .04
Current child’s norms .01(.01) .05
Mother’s past supportive behavior to own parents .18 (.09)* .09
Past affectual/emotional solidarity/closeness -.01 (.01) -.02
Current affectual/emotional solidarity/closeness .09 (.01)*** .44
Past receipt of financial help .41 (.09)*** .21
Inheritance expectation .28(.10)** .12
Child’s gender .39 (.01)*** .19
Child’s income .02 (.01) .08
Child’s education .01 (.03) .02
Child married (vs. not married) -.14 (.10) -.07
Mother’s income .00 (.01) -.02
Mother’s education -.02 (.03) -.02
Mother widowed (vs. not widowed) .02 (.10) .00
Mother’s poor health .06 (.06) .05
123
Table 4.3, Continued
Proximity to Mother in 2000
Past mother’s norms .04 (.03) .07
Current child’s norms -.03 (.03) -.06
Mother’s past supportive behavior to own parents .06 (.19) .02
Past affectual/emotional solidarity/closeness .07 (.03)* .14
Current affectual/emotional solidarity/closeness -.00 (.03) -.02
Past receipt of financial help .42 (.20)* .11
Inheritance expectation .25 (.24) .05
Child’s gender -.08 (.02) -.02
Child’s income -.01 .02 -.03
Child’s education -.07 (.08) -.05
Child married (vs. not married) -.47 (.24)* -.11
Mother’s income -.01 .03 .03
Mother’s education -.14 (.07)* -.11
Mother widowed (vs. not widowed) .40 (.24) .08
Mother’s poor health -.08 (.13) -.03
Note: χ
2
(102) = 254.76, p < .001; NFI = .98; RMSEA = .06
*p < .05. **p < .01. ***p < .001.
124
Table 4.4
Unstandardized, Standardized, and Significance Levels for Structural Equation Model for the
Dependent Variables (Standard Errors in Parentheses); (N = 379)
Instrumental Support Emotional Support
Parameter Estimate Unstandardized Standardized Unstandardized Standardized
Socialization:
Past mother’s norm -.02 (.01)* -.10 -.03 (.02) -.08
Child’s norms 2000 .01 (.01) .06 -.01 (.02) -.02
Social Learning
Mother’s past
support to parents
.24 (.09)** .10 .30 (.14)* .11
Affectual Solidarity
Past closeness .01 (.04) .06 .04 (.02) .11
Current closeness .04 (.01)** .22 .08 (.02)*** .24
Exchange
Past receipt of
financial help
.06 (.09) .03 -.02 (.15) -.01
Inheritance
expectation
.16 (.11) .08 .29 (.16) .09
125
Table 4.4, Continued
Associational solidarity
Frequency of contact -.08 (.05) -.09 .35 (.08)*** .24
Proximity to mother .18 (.02)*** .42 .03 (.04) .04
Characteristics of Child
Child’s gender .15 (.09) .09 .26 (.15) .09
Child’s income -.00 (.01) -.01 .02 (.02) .08
Child’s education -.04 (.03) -.06 -.06 (.05) -.06
Child married -.30 (.11)** -.17 -.26 (.17) -.09
Characteristics of Mother
Mother’s income -.02 (.03) -.08 -.01 (.01) -.02
Mother’s education -.02 (.03) -.03 -.04 (.05) -.04
Mother widowed .19 (.11) .10 .55 (.18)*** .16
Mother’s poor health .12 (.05)* .12 -.02 (.08) .00
Note: χ
2
(102) = 254.76, p < .001; NFI = .98; RMSEA = .06
*p < .05. **p < .01. ***p < .001.
126
Figure 4.3
Graphic Presentations of Significant Results with Standardized Coefficients
Instrumental
Support
2000
Emotional
closeness to
mother
1985
Received financial
support from
parents
1971-1994
Mother demonstrated
support to older
relatives
1971-1985
Emotional
closeness to
mother
2000
1. Social learning
2. Solidarity
3. Exchange and Reciprocity
a. Prepayment
Filial norms
2000
Social /
emotional
support
2000
Contact
with
mother
Widowed
mother
.22
.56
.15
.41
-.17
.21
Mother’s
poor health
.10
.24
.24
.21
Proximity
to mother
.11
daughter
.10
-.14
.18
.11
b. Repayment
Expectation of
inheritance
.09
Mother’s norms
1985
-.10
.44
.12
.19
.14
Mother’s
education
-.11
Married
child
-.11
.11
.12
levels of emotional bonding or solidarity predicted both instrumental and
emotional support. In addition, affectual solidarity in the present also predicted
higher frequency of contact with mothers, which, in turn, further increased the
level of emotional behavior. As expected, strong affective/emotional solidarity
in the past increased the level of such solidarity in the present, thus leading,
indirectly, to higher levels of supportive behaviors towards mothers. When
current affectual solidarity was included in the model, past solidarity did not
127
show a direct affect on actual supportive behavior, yet showed a significant
effect on geographic proximity to mother. Geographic proximity to mother, in
turn, led to higher levels of instrumental support. Interestingly, one other factor
that contributed to higher level of solidarity in the present was the marital status
of the adult child with married children having significantly higher levels of
affectual solidarity.
Exchange models: My analyses provide partial evidence to two of the
proposed exchange models. I found evidence to the “prepayment” model,
although the path was not direct. Adult children who relied on their parents for
financial support at earlier points in their lives were reporting more frequent
contact with their mothers. Higher frequency of contact with mother led to
higher levels of social support to them. This path, therefore, suggests that
mothers may prepay for support in old age by providing financial support to their
children earlier in life as these children will keep more frequent contact with
them and thus provide more support. In correlation analyses, the model
suggests that whether or not a child received such support in the past is
correlated to the level of emotional closeness this child felt to his/her mother
around the same time. This further shows that the role of emotional bonding is
very strong. In addition, the results further showed that adult children who
relied on their parents for financial support at earlier points in their lives were
living in closer geographical proximity to their mothers. Closer geographical
proximity, in turn, led to higher levels of instrumental support.
128
In addition, our results provide evidence to the repayment model as well.
Adult children who expected inheritance were providing higher levels of social
support. This finding suggests that parents may indeed receive support,
specifically social support, in exchange for a future repayment in the form of
bequest.
Associational solidarity as mediating factor: frequency of contact
predicted higher levels of social support and closer proximity predicted higher
level of instrumental support. The results provided support to my mediation
hypotheses, demonstrating that frequency of contact with mother and
geographic proximity to mother play a significant role as mediating variables.
Frequency of contact mediated the relationship between social learning,
affectual solidarity, and the two exchange models—prepayment and
repayment--and social support; Geographic proximity to mothers mediated the
relationship between affectual solidarity and the prepayment exchange model
and instrumental support. It is important to note that frequency of contact was
also predicted by child’s gender with daughters keeping more frequent contact
with mothers. Proximity to mother was further explained by marital status of the
child and education level of the mother with unmarried children and less
educated mothers living in closer proximity.
Characteristics of the adult child and the mother: The study further
emphasized the role of maternal needs and children’s availability. Married
children provided less instrumental support. In addition, as previously
129
discussed, married children lived farther from their parents and thus indirectly
provided less instrumental support to their parents. Mothers in poor health
received more instrumental support and those who were widowed received
more emotional support. Income of both the child and mother and education of
the adult child did not play a significant role. Gender of the child and education
of the mother played only an indirect role. As previously discussed, daughters
were in more frequent contact with their mothers and such contact predicted
higher levels of emotional support. Mothers with lower education lived closer to
their children, thus receiving more instrumental support from them.
E. Discussion
The basic question in this paper was: Can parents influence their
children to provide support to them in old age? The evidence from this study
suggests that the answer is yes, in several ways. Specifically, the results
support three general mechanisms through which parents may shape their adult
children’s supportive behavior including social learning through modeling,
solidarity, and two forms of exchange and reciprocity. Modeling supportive
behavior to one’s own parents yields higher levels of support. Parents may
increase their future receipt of both instrumental and emotional support by
modeling this desired behavior to their children. In agreement with several
previous studies (e.g., Ribar & Wilhelm, 2006; Szinovacz, 1997), adult children
who witnessed their parents provide such support to their grandparents in the
past, demonstrated higher levels of both instrumental and social support to their
130
own parents. Interestingly, witnessing such support did not have a significant
effect on the adult children’s obligation towards providing such support to their
parents, but rather directly on the actual supportive behavior. It further had an
indirect effect on emotional or social support through increasing the frequency
of contact with mothers, which, in turn, predicted higher levels of emotional
support. It is interesting to note that our adult children are reported to have had
witnessed such support anytime between 1971 and 1985. Our adult children
were, on average, as young as 18.6 years old and as old as 32.6 years old
when witnessing such support. The study, therefore, provides evidence to the
powerful role of social learning in young adulthood and the continuous role
parents may have in shaping their adult children’s behavior by modeling desired
behavior to them even when the children may no longer live at home. This is
possibly the most powerful finding of this study as it adds evidence to the
theoretical framework of social learning.
Promoting strong emotional bonds or solidarity with one’s children may
be the most powerful way to promote supportive behavior in old age. Not only
did such bonds predict instrumental and emotional support, but the strength
with which adult children felt close to their parents earlier in life was also
correlated with whether or not the child received a major financial gift from his
or her parents. Receipt of such gift was associated with both contact with and
proximity to mothers, which then led to higher levels of support, suggesting that
such emotional bonds may be the true mechanism behind the exchange
131
mechanism as well. Emotional bonds between adult children and their mothers
are highly consistent across the life course, suggesting that promoting
emotional closeness earlier in life has a powerful effect on intergenerational
relationships later on in life.
As mentioned previously, two mechanisms of life course exchange seem
to operate simultaneously. First, parents may “prepay” for future support by
providing financial support to their children earlier in life. Such downward
transfer from the parents to the children earlier in life led to higher frequency of
contact and closer proximity to mother later in life, which in turn led to higher
levels of social support and instrumental support (respectively). Second,
parents may promise to “repay” for their emotional support by promising a
bequest to their adult children.
My results suggested that mothers’ normative obligations to parental
caregiving, as measured earlier in life when the mothers were of similar age to
that of their adult children in 2000, did not predict the level of normative
obligations of adult children in 2000. Based on the hypotheses of this study,
these results suggest that socialization may not play a role in predicting
supportive behavior. However, I believe that these results should be analyzed
with caution for several reasons. First, it is important to note that these results
are congruent with several previous studies that found weak effects of parents-
children transmission of social norms (e.g., Bengtson, 1975; Glass et al., 1986;
Hoge et al., 1982) as well as weak correlations between parents’ and their
132
children’s values (e.g., Bengtson, 1975; Whitbeck & Gecas, 1988). Additionally,
Miller and Glass (1989), using the first two waves of the same sample as this
current study—the Longitudinal Study of Generations, found that between 1971
and 1985, dyads of children and their parents from the same generations used
in this analysis showed stability in their similarity in several beliefs and values
including religious ideology, political ideology and gender role ideology. In
contrast, dyads from older generations in the study showed divergence in
similarity in those attitudes over the same 14 years. The authors predicted that
the dyads of the generations used in this analysis would diverge and show less
similarity in the future, and that such dissimilarity should, in fact, reflect
successful socialization. Their explanation suggested that successful
socialization would be reflected by divergence in similarity over time, when
other factors, such as peer influence, have more powerful effect on one’s
socialization than his or her parents. Indeed this analysis showed that this
generation does not show significant similarity between 1985 and 2000, in the
specific aspect of filial obligations. Based on Miller’s and Glass’ (1989)
discussion, one cannot conclude whether such dissimilarity indeed reflects
failed or successful socialization as his analysis did not look at the similarity
between the children and the parents earlier in time.
A second and perhaps even more important point to consider when
interpreting these results is based on the first paper of this dissertation. In a
thorough analysis of the malleability of filial norms elsewhere, I have
133
demonstrated that filial norms are malleable to various influences across the life
course. I suggested that filial norms weakened over historical time,
strengthened across successive generations, weakened in response to parental
death and are responsive to other individual variables. Based on this elaborate
analysis, it is reasonable to expect that indeed parental filial norms in 1985
would not predict their children’s norms in 2000 as there might be other
confounding factors affecting these children’s norms. It is, therefore, quite likely
that parents are socializing their children to provide supportive behavior, yet I
was not able to capture this process using the available data. I believe that
future studies should consider other variables to operationalize socialization.
Although not the main focus of this study, the analyses yielded
interesting results with regards to the role filial obligations. Parental filial norms
in 1985, which may be interpreted as parental expectation to receiving support,
were inversely related to actual instrumental support. This result suggests that
parents’ expectations of support from their own children may not be a reliable
predictor of actual support. This result is congruent with several previous
studies (e.g., Eggebeen & Davey, 1998; Lee et al., 1994), that suggested that
supportive behavior on the part of children is not linked to parental
expectations. As far as current filial norms held by adult children, the analyses
showed that adult children who had stronger normative obligations, kept in
more frequent contact with their mothers, and therefore, provided higher levels
of emotional support to them. This result is congruent with other studies, as
134
previously reviewed, suggesting that the strength of filial norms predicts actual
supportive behavior, although this relationship is not direct, but rather mediated
by frequency of contact. Interestingly, however, mothers who self-rated their
health as poorer had adult children with weaker filial obligations. This result
calls for further analyses that take into account other measures of health status
and possibly more objective measures such as health conditions rather than
self-rated health as utilized in this study.
Overall, the results provide strong evidence to the importance of the
intergenerational solidarity model (Bengtson, 1996; Parrott & Bengtson, 1999)
in explaining parental caregiving behavior over the life course. This study
investigated directly and simultaneously the relationship between most of the
dimensions of the model including affectual solidarity--the positive sentiment
toward other members of the family; associational solidarity--the frequency and
patterns of interaction; consensual solidarity--the degree of consistency in
values, attitudes, and opinions through comparing parents attitudes towards
filial norms in the past and those of their children in the present; normative
solidarity —the strength of commitment to perform familial roles and to meet
familial obligations and specifically filial norms; and functional solidarity —the
degree to which family members actually exchange support and specifically
instrumental and social support from adult children to aging parents. While
previous studies addressed some of these dimensions, the strength of this
particular study is in its ability to include all the dimensions in one investigation.
135
It is especially important to note that these analyses provide evidence to
the mediating role associational solidarity plays in predicting supportive
behavior. In accordance with Cicirelli’s (1983) findings, frequency of
communication between adult children and their parents and the geographic
proximity between them played a significant role in mediating various effects,
including emotional closeness, normative obligations, social learning and
exchange, and on actual supportive behaviors. It is further interesting to note
that frequency of communication and geographic proximity each played a
unique role in predicting a different type of support. Geographic proximity
predicted higher levels of instrumental support whereas frequency of
communication predicted higher levels of emotional or social support. These
results provide an important contribution to the literature as they clearly indicate
the differences between these two types of support and validate the choice this
paper took by simultaneously predicting both types of support. Instrumental
tasks such as providing transportation, help with household chores and caring
for a sick parent require the adult child to physically be with the parent, and thus
closer proximity between the two allows for higher levels of such support.
Social-emotional support, on the other hand, including providing information,
advise, emotional support and help with making important life decisions, can be
done from afar, and thus can be achieved using other forms of communication
such as phone conversations, mail and, e-mail communication. The results,
therefore, provide further evidence to the importance of families in spite of
136
geographic distance, or what was termed “intimacy–at-a-distance” (Silverstein &
Bengtson, 1997).
It is important to note that the sample used in this analysis is not a
nationally representative sample. It lacks adequate representation of ethnic and
racial minorities and is a highly educated sample that perhaps represents
mostly the middle class and not lower income families. It is quite likely that the
dynamics described here may be different in lower-income families, where
financial exchanges are less frequent, and in families with different ethnic or
racial cultures. Further studies using more diverse samples are needed to
conclude that the patterns described here are universal mechanisms.
Nevertheless, given the results, I can further conclude that supportive behaviors
towards aging mothers are not only a result of current parent-child relationships,
but are rather shaped through a complex set of factors over the life course.
Parents may in fact shape the supportive behaviors their adult children would
exhibit towards them in old age in more than one way. Adult children will
provide more support to parents they feel close to and felt close to from earlier
points in life, to parents who modeled such parental support to their own
parents–the children’s grandparents—and to parents who provided them with
financial support earlier in life and/or promised them a bequest. While some of
these factors affect the provision of support directly, others determine whether
children live closer to their parents and how frequently they communicate with
137
them, which, in turn, affect the level of support their parents will receive from
them.
138
Chapter Five: Congruence and Discrepancy Between Filial Norms and Filial
Behavior: The Role of Religiosity
A. Introduction
Declines in mortality rates over the past century have increased the need
of older adults for prolonged periods of care, thus making caregiving to older
parents a normative activity in the lives of adult children (Brody, 1981). The role
of adult children in providing long-term support and care to their aging parents
has aroused much interest in social gerontology and family studies in the last
quarter century (e.g., Brody & Brody, 1989; Logan & Spitze, 1995; Silverstein &
Parrott, 2001), yet the norms of filial obligation that guide these supportive
behaviors are not well understood. As a social norm, filial responsibility reflects
the generalized expectation that children should support their older parents in
times of need (Cicirelli, 1988, 1990). More than an expectation of one’s own
behavior, norms of filial responsibility refer to the recognized duties and
obligations that define the social role of adult children with respect to their aging
parents. While previous studies (e.g., Bromley & Blieszner, 1997; Silverstein et
al., 2006; Peek et al., 1998) showed that these expectations are enacted into
actual provision of parental support over time when such support is needed, it is
clear that the relationship between norms and actual supportive behavior is
more complex and is influenced by factors at the individual, familial and societal
levels. While several studies addressed ethnic and racial differential in the
endorsement of filial norms as well as the effects of national contexts on the
139
endorsement of filial norms, what seems to be missing in the literature are
studies that address the role of religious affiliation as well as religiosity in
shaping the strength of filial obligations and in providing a context within which
such normative obligations are enacted into actual supportive behaviors. The
purpose of this study was two-folded. The first aim was to examine the
congruence or discrepancy between: (a) filial norms--what people report they
feel adult children are obligated to do, or should do, in terms of supporting their
aging parents; and (b) filial behavior--what people actually are doing to support
their aging parents. The role of norms or reported intentions in predicting actual
behaviors has long been a controversial topic and a center of debate. This
study specifically examines whether people can be classified into specific
classes or types that describe various patterns of congruence or discrepancy
along these two dimensions of norms or intentions and actual behavior in the
specific topic of filial norm and behavior.
The second aim of the study was to investigate the role of religious
affiliation, level of religiosity, and other individual or contextual factors in
explaining the consistency or discrepancy between reported norms and actions.
In other words, the study examines what would make one more likely to fall into
a specific congruent or discrepant category or type as cross-classified by the
two dimensions of filial norms and filial behavior. This study specifically aimed
at filling the almost complete absence of studies on the specific role of religion
140
and religious affiliation as contextual factors in understanding the development
of filial norms and their enactment into filial behavior.
Religion, Religiosity and Families
Religion and religiosity have recently attracted the attention of
sociologists and family scholars, who are uncovering the role religion and
religiosity play in several domains in individual and family life. For example,
several studies (e.g., George, Larson, Koenig & McCullough, 2000; Koenig,
McCullough & Larson, 2001) addressed the role of religious behavior in
contributing to positive physical health outcomes and better psychological well-
being. Other studies have addressed the mechanisms through which religion
and religiosity may be transmitted across generations (e.g., Copen, Biblarz,
Silverstein, & Bengtson, 2005), suggesting that grandmothers have a unique
influence in the grandchild’s religiosity independent of that of a parent. Yet
others investigated the role religion and religiosity play in forming family unions
and specifically cohabitations versus marriages. For example, Thornton, Axinn
and Hill (1998) showed that religious affiliation plays a role in entering a family
union. Specifically, Catholic women had the lowest rates of entering unions—
both marriage and cohabitations—and individuals with no religion affiliation had
a risk of entering cohabitation that was 50% greater than other affiliated
individuals. They additionally showed that various religiosity dimensions and
specifically frequency of attendance and importance of religion were negatively
141
related to cohabitation, especially for women, and that maternal frequency of
religious attendance reduced rates of cohabitation.
More specifically to the topic at hand, there have been several studies
that addressed the role religion and religiosity may play in affecting the various
relationships between family members. For example, Call and Heaton (1997),
using the National Survey of Families and Households, found that the frequency
of religious attendance had a positive effect on marital stability and that
incongruence between spouses in church attendance increases the risk of
divorce. They, however, did not find significant differences in marriage stability
based on religion affiliation. Other studies specifically addressed the child-
parent relationship. For example, using the Intergenerational Panel Study of
Mothers and Children, Pearce and Axinn (1998) concluded that while religious
affiliation did not have a significant effect on parent-child relationship, religiosity
had a significant positive effect on both mothers’ and children’s reports on their
quality of their relationships. Specifically, they suggested that personal
importance of religion was more important than the more public domain of
religiosity, measured by religious participation, in their effect on the mother-child
relationship. This study is of special importance to this investigation as it
suggests that religiosity plays a role in strengthening the affective relationships
between children and parents. Based on the previous chapter as well as
previous studies (e.g., Cicirelli, 1983; Rossi & Rossi, 1990; Silverstein et al.,
2002), it is clear that strong affective ties between adult children and their
142
parents are a strong predictor of support provided to parents when such support
is needed. In fact, the previous chapter suggested that affectual ties were the
strongest predictor of support. It is, therefore, important to investigate factors
that might strengthen such parent-child ties. Religion and religiosity seem to be
such important factors that are unfortunately largely overlooked by most studies
on family relationships and intergenerational exchanges of support (Pearce &
Axinn, 1998).
Religion, religiosity and parental care
In spite of the growing number of studies on the link between the study of
religion and study of families, and the evidence that religion and religiosity may
play a role on parent-child relationship, an extensive literature review suggested
that there was only one empirical study that directly addressed the role religion
may play in intergenerational assistance (Myers, 2004). Myers (2004)
concluded that the relationship between religion and religiosity and
intergenerational exchanges of support was quite complex. Myers (2004) found
that the relationship between religiosity and parental assistance was
conditioned on gender of the child and the marital status of the parent.
Specifically, religiosity was negatively associated with support levels given to
divorced fathers. Religiosity predicted higher levels of support from adult
daughters to continuously married parents. Additionally, religious congruence
between adult children and parents predicted higher levels of support. His most
consistent finding showed that intergenerational congruence in Evangelical
143
Protestantism was associated with lower levels of assistance from adult sons to
divorced mothers and from adult daughters to divorced fathers.
Religion and Religiosity as Crucial Contexts in Understanding Filial Norms and
Behaviors
The current focus on the role of religion and religiosity as crucial contexts
within which filial norms develop and are enacted to supportive behavior is
based on three general notions.
First, the dictate of the fifth commandment to honor one’s father and
mother is held by all western religions, and the filial responsibility for the care
and support of elderly parents is a hallmark of almost all societies and cultures
in the world. However, based on previous studies, it is clear that there are
variations among social groups in the strength with which these values are held
and are enacted (see literature review section for detailed discussion on ethnic,
racial, and cross-national differences). Individuals who are more involved in
practicing religion through both the public sphere (participation in religious
services) and the private spheres (family prayers, personal salience of religion)
are more likely to be in more frequent contact with the various religious
messages that promote strong family commitment (Pearce & Axinn, 1998). It is
only logical to suggest that the norms of filial obligations as well as the actual
filial behavior may be influenced by religious affiliation as well as by level of
religiosity.
144
Second, religion as a social network is argued to be protective in that it
promotes exchange of support with other individuals. Because religion is
usually based on values of collectivism, religious individuals are expected to
provide support to other individuals (Myers, 2004). Religious organizations
often are involved in providing formal forms of support to needy individuals. For
example, in Judaism, the concept of “Tikkun Olam” or making the world a better
place, involves constant efforts to support the needy by providing meals to the
homeless, toys and books to needy families etc. Synagogues, Jewish day
schools and religious schools represent a natural place where such projects are
organized. In fact, in Judaism, the efforts to provide support to the needy are
organized in a hierarchical order with specific order of first helping the needy in
one’s own community, those in closer proximity to one’s community and finally
those who are farther away. There are also certain family conditions that put
one at a priority for care including the orphans and the widowed. It is likely to
suggest that individuals who are more involved in such projects through
participation in religious ceremonies that stress these ideas, as well as actual
projects, may be more inclined to provide any type of assistance to other
individuals in both formal and informal fashions, and as a result more likely to
provide support to his/her own parents in times of need.
Third, most religious groups endorse strong family relationships (Myers,
2004). Religious institutions typically provide opportunities for family activities
including family camps and retreats, which, in turn, may promote stronger
145
intergenerational family bonds (Pearce & Axinn, 1998). Such family
involvement may suggest that the religious commitment to helping other family
members may be even stronger than the commitment to help other individuals.
In addition, it is likely that various religious affiliations may vary with
regards to their views and beliefs with regards to family relations in general,
adult child-parent relationship specifically, and even more specifically with
regards to the specific aspect of filial obligations and filial behavior (Heaton &
Pratt, 1990; Pearce & Axinn, 1998). Taken together, these three notions, along
with the empirical evidence that supports the role religion and religiosity may
play in various aspects of parent-child relationships and potentially in parental
caregiving situations, suggest that religion and religiosity should be studied as
possible predictors of such parental support.
Other Factors Affecting Filial Norms and Behavior
The literature review of this dissertation provided an extensive review of
the various factors that may influence both the development of norms and the
enactment of such norms into actual behavior. Because the current study
focused on congruence or discrepancy between what one feels obligated to do
and what one actually does, four important concepts were considered. First, it
is important to consider geographic proximity to parent. The second paper of
this dissertation suggested that geographic proximity played an important role in
the amount of instrumental support one may be able to give to his/her parents.
It is also likely that while adult children may develop strong filial obligations to
146
their parents, they might not act upon those obligations simply because of lack
of opportunity as a result of the physical limitation of living far from one’s
parents. The second important factor is the role of possible competing
demands. Here too, it could be that adult children may develop strong
normative expectations, yet might not be able to deliver the actual support
because of competing responsibilities. In this analysis, the focus is on
competing norms in the form of having children.
The third factor involves the degree of similarity between adult children
and parents on values, opinions and life views. This aspect is of particular
importance to this analysis. The above described previous studies on the
association between religion and families (e.g., Call and Heaton, 1997; Myers,
2004; Pearce & Axinn, 1998) suggested that while individual religiosity plays a
role in predicting the quality of family relationships, the congruence in religiosity
between parents and children has a powerful predicting power as well.
However, previous studies using the sample used in this current study (e.g.,
Roberts & Bengtson, 1999; Bengtson & Roberts, 1991) did not find that value
congruence was consistently related to exchange of support. Value
congruence was, therefore introduced in this analysis separately from individual
religiosity. Finally, the fourth factor in this analysis is an alternative caretaker in
the family. It was suggested that the structure and gender of sibling sets may
affect the level of provision of care by adult children to their aging parents.
Specifically, Wolf, Freedman and Soldo (1997) showed that siblings have a
147
tendency to cut back on their caregiving activities in association with the
number of sisters they have.
B. Research Questions and Expectations
The current investigation has two main areas of focus. The first area of
focus proposes that obligations to parental support, or filial norms, may or may
not accompany actual provision of support and vice versa—actual caregiving
may or may not be accompanied by feelings of obligations. As a result, there
could be various patterns of congruence or discrepancy between norms
regarding parental support and actual supportive behavior and these congruent
or discrepant patterns may be classified into groups. Classifications have been
used to describe diversity in structures and functions of family relations (e.g.,
Cherlin & Furstenberg, 1986; Giarrusso, Silverstein, Gans & Bengtson, 2005;
Hogan, Eggebeen, & Clogg, 1993; Silverstein & Bengtson, 1997; Silverstein &
Litwak, 1993). Our research questions and expectations in this first area of
focus are as follows:
1. How many classes are needed to adequately describe the possible patterns
formed by classifying individuals based on two dimensions — norms and
behavior – in our data? How many classes would there be in the best-fitting
model?
While mathematically, there could be more options (2 to the power of 6),
theoretically we were interested in distinguishing among the congruent and the
discrepant factors. Figure 5.1 describes the four possibilities of those
148
discrepant or congruent factors. The two congruent factors that may arise are
as follows: those who show high filial norms and high actual provision of
support. This class would be termed committed supporters, as they both show
high commitment and high actual supportive behavior; and those who show low
filial norms and low actual provision of support. This class would be termed
uninvolved. The two discrepant patterns would include: those who report high
filial norms yet provide little support. This class was termed dissonant; and
those who report lower levels of commitment yet provide higher levels of
support. This class would be termed doers, as they provide the support yet lack
the supposed obligatory motivation.
2. What are the profiles of classes in the best fitting model? In other words,
what are the probabilities and latent class distributions for the constrained
best-fitting model?
3. Are the patterns of congruence and discrepancy based on filial norms and
filial obligations the same class-types across various religious affiliations?
Specifically, the paper investigates the following groups: Jewish, Catholic,
Protestant, No Affiliation, and Other. This investigation was exploratory in that it
did not attempt to test specific hypotheses on the expected differences in the
distribution of these class-types across various religious affiliations. However, it
is likely that there might be differences across various religious affiliations. In
addition, it is likely that individuals who have no affiliation may have less
149
Figure 5.1
Hypothetical Classes of Congruence and Discrepancy Based on the Classification of Individuals
on the Two Dimensions of Filial Norms and Filial Behavior
Actual Support
Low High
Low Predictably not
committed
Uninvolved
Surprisingly caring
Doers
Filial
Norms
High Do not act upon
convictions
Dissonant
Reliably committed
Committed
Supporters
exposure to the various activities and beliefs common to all religions that
support familism and caring. If this is the case, it is likely that those non
affiliated individuals would be less likely to develop norms or to act upon such
norms and as a result, one might expect higher percentages of uninvolved or
dissonant in this group.
The first focus of analysis is descriptive in nature. However, this
investigation attempted to go beyond the mere description of the class-types by
attempting to explain what characteristics may place one at higher risk of being
150
in one group versus another. The second focus of this paper, then, was to
address the possible explanations of these type-classes and specifically to ask:
4. Can one predict who was more likely to fall into a specific class-type?
More specifically, I posed several questions with regards to the specific
variables at the center of analysis. The most important question had to do with
the main focus of this analysis with regards to religiosity. Based on the
literature review in this paper it appears that exposure to religion in any form is
likely to increase one’s likelihood of caring for others and especially for family
members. It is important, therefore, to look for the extreme case of those who
are non-religious and are not exposed to religion. I am specifically asking,
therefore, are non religious people more likely to be dissonant or uninvolved?
That is, to either develop normative obligations towards the care of aging
parents and not act upon these convictions, or to even not develop such
obligations?
The paper further posed several other questions with regards to other
variables of interest. Specifically, are people who live far or those who have
competing demands more likely to be dissonant? And are adult children who
feel similar to their parents more likely to be committed supporters?
C. Methods
Sample
The sub sample for this analysis is comprised of 475 subjects from the
Longitudinal Study of Generations, who had at least one surviving parent that
151
exhibited some difficulty in at least one activity of daily living (ADL) in 2000. It
was crucial to limit the involvement of subjects to those with surviving parents
who exhibited some level of need. If there were no need on the part of the
parents, the paper could not have assumed that the possible discrepancy
between norms and actual support was a result of any other factor, but rather a
mere reflection of the fact that the adult child was not called upon to provide the
relevant aid. Table 5.1 provides a description of the sample. The sample’ s
mean age was 51.26, with 58.6% married individuals. Table 5.2 represents the
distribution of subjects by religious affiliation.
Measures
Filial norms and filial behavior: In an attempt to address the discrepancy or
congruence in the most accurate way, I chose two variables that represent each
of these concepts — norms vs. behavior -- on the same dimension.
Filial norms: Two items were chosen for this analysis. Both items are part of a
section in the questionnaire that asks subjects about their opinions on social
and family issues. Respondents were instructed to indicate how much they
agree or disagree with each statement on a 4-category Likert scale ranging
from strongly disagree to strongly agree. The relevant items are as follows: (1)
“As many activities as possible should be shared by married children and their
aging parents”; and (2) “It is the responsibility of adult children to be with their
parents in time of serious illness even if the children have moved some distance
152
Table 5.1
Descriptive Statistics (N=475)
Variable %
Number Mean (SD)
Age
51.26 (8.65)
Gender (Female) 57.5 273
Education (Some College +) 73.3 348
Marital Status (Married) 79.8 379
Has children 72 342
Has sister 67.2 319
Not Religious 14.7 70
Live farther than 50 miles 16.6 79
Similarity to parents 3.6 (1.2)
Table 5.2
Description of Sample by Religious Affiliation
Characteristic % Number
Jewish 11.4 54
Catholic 13.3 63
Mainline Protestant 21.1 100
No Affiliation 28.8 137
Other 25.6 121
153
from their parents”. For each item, responses were dichotomized whereby 1
represents agree or strongly agree with the statement, reflecting high level of
filial norms, and 0 represents disagree or strongly disagree with statement,
reflecting low levels of filial norms.
Filial behavior: Two corresponding items were chosen for this analysis. Both
items are part of a section on exchange of help and support, where
respondents are instructed to specify the frequency with which they provide a
specific type of support to their mother and father. The two specific items
chosen for this analysis are: (1) “visiting/sharing leisure activities”; and (2) “help
when she/he is sick”. Respondents are asked to respond to this question using
an 8-category Likert scale ranging from “not at all” to “daily”. For those with one
parent surviving, the frequency of providing support to that parent was included
in the analysis. For those with two surviving parents, the analysis included
information about the parent for which the adult child provides the highest
amount of support. For the visiting and sharing social activities, the item was
dichotomized, whereby 1 represents visiting one’s parent(s) several times or
more, reflecting high level of filial behavior and 0 represents visiting one’s
parent monthly or less often including not at all) representing low level of filial
behavior. For help when parent is sick, responses were dichotomized whereby
1 represents any frequency of help provided (ranging from once a year to daily)
and 0 represents no provision of any help when sick.
154
Religion and religiosity: The Longitudinal Study of Generations includes a
number of measures to study religion and religiosity, which fall under the
theoretical view of the construct of religion and religiosity as a multidimensional
construct. The following theoretical dimensions are addressed: (1) religious
affiliation using the following item “What is your current religious affiliation?
(Please specify the church or denomination. Write “none” if you have no
affiliation.)”; (2) practice participation in religious activities or public religiosity
using the following item: “How often do you attend religious services these
days” with a 6-category Likert scale ranging from “never” to “more than once a
week”; (3) identity or self-rated religiosity using he following item: “regardless of
whether you attend religious services, do you consider yourself to be” and the
respondent is given four options including: “not at all religious”; “somewhat
religious”, “moderately religious” or “very religious”; And (4) religious idealism
including the level of agreement with the following statements: “All people alive
today are descendents of Adam and Eve”; “This country would be better off if
religion had a greater influence on daily life”; “Every child should have religious
instruction”; “God exists in the form described in the bible”; and “Religion is the
most important influence in my life”. Please see the next section for more
detail.
Although theoretically these aspects of religion and religiosity describe
different dimensions of the multifaceted construct of religion and religiosity,
factor analysis using all these measures (excluding religious affiliation) yielded
155
a single factor. These results suggested that religiosity, as measured in the
LSOG in the specific sub sample of this analysis, represent a unidimensional
construct. As a result, I chose to use two sets of variables--the first to measure
religious affiliation and the second to represent religiosity.
Religious affiliation: Respondents were divided into religious affiliations based
on their answers to the following open-ended question: “What is your current
religious affiliation? (Please specify the church or denomination. Write “none” if
you have no affiliation.)”. Respondents’ responses are coded regularly into
various groups. For the sake of this analysis the following categories were
created: Catholics, Mainline Protestants for those responding any Protestant
affiliations that are not part of the evangelical Protestants; Jewish; and “No
affiliation” for those specifically responding “none” or “no affiliation”. All other
categories along with missing case were clustered under the title “Other”. It is
important to note that this cluster includes various types of denominations
including Mormons, Evangelical Protestants as well as other denomination
which might be very different from each other. As a result, this analysis refrains
from interpreting any results with regards to this category. Religious affiliation is
included as a covariate in the LCA and is used by the software to distribute
individuals across the various class-types.
Non-religiosity: The measures I chose for this analysis are the items related to
religious idealism. All items are part of a section in the questionnaire that asks
subjects about their opinions on social and family issues. Respondents were
156
instructed to indicate how much they agree or disagree with each statement on
a 4-category Likert scale ranging from strongly disagree to strongly agree. The
statements include the following: (1)“All people alive today are descendents of
Adam and Eve”; (2) “This country would be better off if religion had a greater
influence on daily life”; (3) “Every child should have religious instruction”; (4)
“God exists in the form described in the bible”; and (5) “Religion is the most
important influence in my life”. In order to capture the extreme nonreligious
respondents, nonreligious for the sake of this study is operationalized as a
subject who responded he/she strongly disagrees with at least one of these
items.
Other Predictor Variables: Geographic proximity was determined based on a
continuous variable indicating the distance in miles from the child’s residence to
the mother’s residence on a 7-point scale ranging from “more than 500 miles” to
“we live together”. The item was then dichotomized, whereby 1 indicates those
living far (over 50 miles away) and 0 represents those living 50 miles or under
away from their parents. If only one parent was alive, the response regarding
that parent was included. If both parents were alive, the response that
indicated the closest proximity was included. Competing demands were
measured by whether the adult child had or did not have children. Having other
alternative caretakers was measured by whether the adult child had at least one
sister. Finally, similarity between child and parent in values or beliefs was
measured by one item asking individuals on a 6-category Likert scale ranging
157
from “not at all similar” to “extremely similar” “In general, HOW SIMILAR are
your opinions and values about life to those of your parent at this point in time”.
Respondents answered this question with regards to father and mother. If only
one parent was alive, the response regarding that parent was included. If both
parents were alive, the response that indicated higher similarity was included.
In addition, the analysis controlled for age of the respondent, education
of the respondent (1 = some college and above vs. 0 = less educated), and
marital status (1 = married vs. 0 = all other).
Procedure
The first three research questions in this paper, representing the first
focus of this paper, require the use of Latent Class Analysis (LCA). LCA is a
statistical method that allows researchers to propose that a set of unobserved
or latent class-types account for the association among cross-classified
categorical variables (Clogg, 1995; Clogg & Goodman, 1984; Giarrusso et al.,
2005; Silverstein & Bengtson, 1997). This analysis suggests that the
discrepancy and or congruence between filial norms and filial behavior can be
parsimoniously described by a set of typical classes that are empirically
manifested by cross-classification of several measured dichotomized variables.
A key assumption in LCA is that the membership in a latent class is the true
source of covariation among the measured variables. As a result, a set of latent
classes is acceptable to the extent that it minimizes the within-class association
among the observed variables, and the statistical test allows the researchers to
158
decide whether a theoretical model adequately describes the observed data
(Giarrusso, et al., 2005; Silverstein & Bengtson, 1997). Latent Gold (Vermunt &
Magidson, 2000) was used to estimate models. It is important to note that the
goal of such typology is to identify a parsimonious model that allows to
adequately describe the various observed possibilities in a manageable and
accurate number of clusters.
When estimating the models, two sets of parameters should be
considered; (1) the adequacy of the model. As a first step, the adequacy of the
model is tested yielding a likelihood chi-square test (L
2
), which tests for
statistically significant discrepancies between a theoretical model and the
observed data. The likelihood chi-square test should be nonsignificant
indicating no significant differences between the theoretical model and the
observed data. In addition to the likelihood chi0square statistic, one should also
consider the Baysian Informal Criterion (BIC) statistic, which becomes useful
when selecting the best-fitting model among competing ones based on various
numbers of clusters. The best-fitting model is that with the lowest BIC (Clogg,
1995; Raftery, 1986). Additionally, one may also choose to look at the
classification error statistic, which indicates the cases that were misallocated in
the specific model. The lowest number of classification errors is preferable;
And (2) the parameters of the model. Two parameters are estimated. The first
is conditional probabilities. These probabilities reflect the distribution of the
observed indicators for members of each latent class. These are similar to
159
factos loadings in that they represent the association between observed and
latent variables. The second is latent class probabilities, which signify the
distribution of members across types, and thus allows the researcher to
describe the prevalence of type in a population (Giarrusso et al., 2005;
Silverstein & Bengtson, 1997).
Once the model is selected and the profiles of each class-type is
described, the fourth research question, which represents the second area of
focus of this paper is addressed using multinomial logistic regression analysis.
The dependent variable at this point became class-types and thus was a
categorical variable with three categories: (1) dissonant; (2) uninvolved; and (3)
committed supporters. Multinomial logistic regression is the appropriate
method for predicting such dependent variable as it is a generalized linear
model that extends linear regressions to noncontinuous outcomes (Crosnoe,
Mistry, & Elder, 2002). Because respondents in this sub-sample may be
members of the same family, robust standard errors are calculated to adjust for
family dependencies such as parent-child, siblings, cousins, or uncle/aunt-
nephew/niece. For this paper, several variables were included in the analysis
including: being non-religious, living far from one’s parents, having children,
having a sister and having similar values to one’s parents. Several socio-
demographic variables were included in the analysis as control variables
including gender, education and marital status. The category of committed
supporters was chosen as the reference group and the analysis, therefore,
160
yields results that allow the researcher to decide what are the odds of a
respondent falling into either the dissonant or the uninvolved group in reference
to the committed supporters group.
D. Results
Latent Class Analysis:
Table 5.3 represents the chi-square, BIC and classification error statistics for
each model. Figure 5.2 and 5.3 represent a graphic presentation of the BIC
and classification error statistics respectively. The first model is the one class
model, which assumes that there are no relationships between the four
measured items. This model fits the data poorly based on the chi square
statistic. The two-, three-, four-, and five-class models, and marginally the six-
class model have a good fit based on the chi-square statistic. When choosing
between the competing models, I first looked at the BIC statistic. This statistic
is lowest for the two-class model and then goes up with each additional added
class. However, the classification error for the two- and three-class models is
very similar, suggesting that adding one more class does not lead to more
classification errors. It is suggested, therefore, that the three-class model is the
best-fitting model for the data. Once the best-fitting model was selected, I
examined and interpreted the three-class model. Table 5.4 represents a
summary of the latent class conditional probabilities and class distributions.
The conditional probabilities describe the profile of each class. Figures 5.4, 5.5,
161
and 5.6 represent graphic presentations of the profiles for each class-type. The
latent probabilities describe the class distribution.
Table 5.3
Latent class models of intergenerational relationships using four dichotomous indictors of filial
behavior and norms
Models L
2
df P BIC Classification
error
One class –
complete independence
153.75
71
4.8e-8 -268.75 0.00
Two class 71.76 66 .29 -320.99 0.16
Three-class 62.84 61 .41 -300.15 0.17
Four-class 61.32 56 .29 -271.91 0.31
Five-class 61.31 51 .15 -242.17 0.45
Six-class 61.22 46 .06 -212.51 0.44
Seven-class 61.23 41 .02 -182.75 0.53
Eight-class 61.32 36 .005 -152.90 0.58
162
Table 5.4
Latent Class Probabilities and latent class distributions for constrained three-class model
Latent class probabilities for a three-classes
Measure Dissonant Uninvolved Committed
supporters
Filial Norms
Should share activities .36 .56 .65
Should care when sick .34 .99 .80
Filial Behavior
Visit frequently .02 .30 .90
Care when ill .48 .04 .97
Class Distribution .52 .24 .24
Figure 5.2
BIC statistic for models with 1 through 8 classes
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
1 2 34 567 8
Number of Classification
Classification Error
163
Figure 5.3
Classification error statistic for models with 1 through 8 classes
-350
-300
-250
-200
-150
-100
-50
0
12 345 6 7 8
Number of Classes
BIC
The results suggest only three, rather than the hypothesized four class-
types. About half of the sample falls under the category of dissonant, as they
report that one should share activities and should care for parents when ill, yet
did not provide frequent support or frequent visits. About a quarter of the
remaining sample fall under the category of uninvolved. These subjects are
likely to report low levels of both norms and behavior. Finally, the last quarter of
the sample fall under the category of committed supporters, suggesting that
these individuals are likely to both state strong filial norms and actually provide
equivalently high levels of such support. There is no indication of a category of
164
doers—those who would exhibit high levels of support yet without the
accompanying normative obligations.
The next step in this specific analysis is to examine the distribution of
these three class-types across religious affiliation. Religious affiliation was
included in the cluster analysis as a covariate, and the clusters were drawn
using the information on religious affiliation. Figure 5.7 represents a graphic
presentation of the distribution of the classes across religious affiliation. While
these differences were not tested for significance, one can interpret the results
in terms of possible trends. Specifically, it appears that among Catholics, the
proportion of individuals who fall into the committed supporters in the highest
Figure 5.4
Latent class conditional probabilities for class 1: Dissonant
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Should Share
Activites
Should Care in
Sickness
Visit Often Support When
Sick
Norms-Activities Items
Conditional Probablities
165
Figure 5.5
Latent class conditional probabilities for class 2: Uninvolved
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Should Share
Activites
Should Care in
Sickness
Visit Often Support When
Sick
Norms - Support Items
Conditional Probablities
Figure 5.6
Latent class conditional probabilities for class 3: Committed Supporters
0
0.2
0.4
0.6
0.8
1
1.2
Should Share
Activites
Should Care in
Sickness
Visit Often Support When
Sick
Norms -Support Items
Conditional Probablities
166
compared to other religious groups. Jewish and Mainline Protestants seem to
share a similar high proportion of respondents categorized as dissonant. The
non-affiliated indeed have a high proportion of uninvolved, as expected, yet this
proportion is very similar to that of uninvolved among the Jewish group.
Figure 5.7
Distribution of the three class-types by religious affiliation
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Jewish Catholic Mainline
Protestant
No Affiliation
Religious affiliation
Committed
Supporters
Uninvolved
Dissonant
Multinomial Logistic Regression:
The results of the multinomial logistic regression are presented in Table
5.5. Figure 5.8 represents the significant results in a graphic presentation. The
167
analyses were run using the third group--committed supporters—as a reference
category, suggesting that the results should be interpreted as the risk of one’s
being in the dissonant or uninvolved group in reference to the committed
supporters groups. A significant odds ratio with a value below 1 indicates that
the independent variable reduces the odds of the dependent variable having a
value of 1 (being in the dissonant or the uninvolved group), and an odds ratio
greater than 1 indicates an increase in these odds. In Table 5.5, the left column
represents the coefficients for being in the dissonant group vs. the committed
supporters group, and the right column represents the coeifficients for being in
the uninvolved group as opposed the committed supporters group.
Two variables significantly predicted membership in the dissonant group.
Those who lived far from their parents were about 42 times more likely to be
dissonant than to be committed supporters. In addition, those who had children
were about 5 times more likely to fall in that group. Two variables significantly
predicted membership in the uninvolved group, and one was close to
significance level. Those living far were more likely (about 50 times more likely)
to be uninvolved than those who lived close. For the uninvolved group, an
additional variable was a significant predictor. Specifically, a non-religious
individual was about three times more likely to fall into the uninvolved group as
opposed to the committed supporters group. While not reaching significance
level in this group, having children was very close to significant predictor in
increasing the risk that one would be uninvolved versus committed.
168
Table 5.5
Summary of multinomial logistic regression analysis for variables predicting class-type affiliation
for dissonant and uninvolved in reference to committed supporters
Dissonant Uninvolved
Predictor e
B
Robust
SE
e
B
Robust
SE
Non religious 1.39 .59 2.82* 1.36
Live far from parent 42.11*** .31 50.79*** 40.57
Have children 4.66* 3.51 4.92
†
4.05
Have sister 1.04 .46 1.13 .52
Similar views to parent 1.00 .17 1.01 .64
Female .75 .32 .94 .49
Some College Education 1.21 .58 1.65 .97
Married .75 .46 .81 .52
Pseudo R
2
.164
χ
2
(df)
55.84 (16) ***
Note: Reference groups is strongly committed
†
<.10, *p < .05. **p < .01. ***p < .001.
169
Figure 5.8
Log Odds ratios for membership in the dissonant or uninvolved class-types versus the
committed supporters class-type
0
10
20
30
40
50
60
Not Religious Lives Far Has children
Characteristic
Odds Ratios
Dissonant Uninvolved
Note: the reference group is committed supporters.
E. Discussion
This paper had two main areas of focus. The first area of focus proposed
that there might be patterns of congruence as well discrepancy between the
sense of obligation people have for parental care and what people actually do in
terms of parental support. In other words, the paper proposed that normative
obligations may not always accompany parental supportive behavior and vice
versa, supportive behavior may not always follow normative obligations. The
170
data showed that there were three patterns of cross-classification between the
two items measuring commitment to care and the two items measuring actual
supportive behavior. Two of the patterns are congruent and one was
discrepant. Interestingly, the most prevalent pattern in this sub-sample, held by
about half the respondents, was a discrepant pattern, where individuals
reported strong commitment to care, yet did not provide much care. This type
of pattern was termed dissonant. The second most prevalent was a congruent
pattern, where individuals had weak commitment to care and not surprisingly,
did not report providing the relevant care. This type was termed uninvolved.
The third pattern, which was only slightly less prevalent than the second, was
the opposite pattern, those who had strong commitment and provided much
support. This congruent pattern was termed committed supporters.
The second area of focus of this paper examined the role religious
affiliation and religiosity may play in placing one in one group versus another.
Specifically, it examined the risk of one falling into the discrepant or uninvolved
group versus being a committed supporter. The paper suggested that the
distribution of these three patterns may be different across religious affiliations.
As was the case with previous studies (e.g., Pearce & Axinn, 1998), this
analysis did not show any significant differences by affiliation. However, while
not significant, several results are worth mentioning. Specifically, it appears
that among Catholics, the proportion of individuals who fall into the committed
supporters is the highest compared to other religious groups. Jewish and
171
Mainline Protestants seem to share a similar high proportion of respondents
categorized as dissonant. The non-affiliated indeed have a high proportion of
uninvolved, as expected, yet this proportion is very similar to that of uninvolved
among the Jewish group.
It is interesting to examine these results in conjunction with the next set
of results with regards to non-religiosity. Non-religious people seemed to be
significantly more likely to fall in the uninvolved group rather than to be
committed supporters. Non-religiosity may be related to both the development
of filial commitment and the actual delivery of parental support. In other words,
religiosity may in fact, as suggested in this paper, be an influential context
within which filial norms develop and are enacted to actual supportive behavior.
This paper proposed that religion and religiosity may influence the development
of filial norms and the enactment of such norms based on three premises. First,
most religions support showing honor to one’s parents as dictated by the Fifth
Commandment. Secondly, religious activities usually instill values of
collectivism that encourage caring behaviors to others and provide a context
that promotes both informal and formal venues for helping others. And finally,
religious activities usually promote a strong sense of family cohesiveness and
commitment, theoretically leading to higher levels of commitment and support to
one’s family members. Although this paper does not provide a specific answer
that would indicate the specific mechanism through which the effects of religion
and religiosity influence the development and enactment of filial norms into
172
actual filial behavior, it provides evidence that such link may truly exist, and
opens the door for future investigations to attempt a more specific explanation.
When attempting to interpret these results, it is important to note that
several other affiliations were not specifically included in this analysis. Most
notably, various groups such as evangelical Protestants and Mormons were all
included in the ‘other” category. It is likely that these groups are different from
each other as well as from the various affiliations tested in this analysis. For
example, Evangelical Protestants represent a unique group that may strive for
more cohesive intergenerational relationships as they promote a unique pro-
family culture (Myers, 2004; Wilcox, 1998). While this specific sample did not
have a large enough group of respondents belonging to each of these groups to
provide for a meaningful analysis, the upcoming new wave of the LOSG may
provide such sample. Based on the results reported here, it is my belief that
one should consider religious affiliation and pay special attention to specific
groups that might be different from others in promoting both filial commitment
and filial behavior. I also believe that this paper’s results suggest that
researchers should be encouraged to look at religion affiliation in conjunction
with religiosity in future analyses. For example, Miller and Stark (2002) discuss
religion and religiosity in the context of religions with high versus low risk. They
suggest that some religions are considered high-risk religions, where the
consequences of behaving in an irreligious way are believed to be severe.
Some of these high risk religions believe that nonreligious people will receive
173
divine punishment possibly in the next world. This view may imply that one may
be less likely to behave in a non-religious way in a higher risk religion.
Combined with this paper’s proposition that filial norms and behaviors may be
promoted by and viewed as religious behavior, one may speculate that among
Christians, there would be more committed supporters, and among the non-
affiliated, there would be more uninvolved individuals. Although not significant,
the results in this study show such exact trends. It is also not surprising that the
Jewish population in this sample was not much different from the un-affiliated
as, based on previous analysis, the majority of the Jewish population in this
sample is secular.
Finally, this paper confirms other studies, which suggest that geographic
proximity, or opportunity, along with competing demands, in the form of
children, play a significant role in one’s ability to provide support to parents.
This study shows that opportunity, or geographic distance from one’s parents, is
the most important predictor of the pattern of congruence or discrepancy
between filial commitment and behavior. Individuals who live far from their
parents, are about forty times more likely to be dissonant, that is to not be able
to act upon their convictions, showing that although they are committed to care
for their aging parents, the geographic distance between them, make it less
likely that they would actually provide that care. As expected, having children
as a competing demand also plays a role in predicting whether one will act
upon one’s convictions and provide support. Interestingly, however, the
174
findings also suggested that both opportunity—geographic proximity and
possibly competing demands—having children (although that variable did not
reach significance level), increased one’s risk of being uninvolved, that is not to
report higher filial commitment. While on the surface, it might suggest that
people who live far or have competing demands may not develop filial
commitment, there might be a more insightful explanation. Reporting strong
filial norms for those who are unable to provide support puts them at a state of
dissonance. When discussing the differences between generalized filial norms,
as measured in this paper, and more specific personal intentions—what people
say that actually plan to do with their own parents, Finley et al. (1988)
suggested that generalized filial expectations may be adjusted in an attempt to
reconcile the gap between the ideal and what is possible or actual. In fact, they
specifically discussed people whose competing demands or geographic
placement prevents them from providing the support they would have liked to
provide. Applying these suggestions to this current investigation, it is likely that
some of the respondents in the uninvolved report low filial commitment as a
way of reconciling the difference between the ideal form of care they would
have liked to provide to their parents, and what they actually can provide, which
is lower due to geographic distance, competing demands or other forms of
barriers to actual support provision.
Religion continues to be an important influence in American life (e.g.,
Smith, 2005). This paper proposed that religion and religiosity are likely to
175
affect filial norms and filial behavior. The analyses supported such proposition
and provided evidence to suggest that indeed religion and religiosity are
influential contexts within which filial norms are developed and are enacted into
actual supportive behavior. The analyses further showed that the cross-
classification of normative obligations towards parental care and the actual
provision of support may yield both congruent and incongruent patterns
suggesting that the relationship between filial norms and filial behavior is a
complex relationship that should continuously be studied within the multiple
contexts within which such support is needed and provided.
176
Chapter 6: Discussion and Conclusions
A. Summary
Each of the three papers in this dissertation have made a meaningful
contribution to our knowledge about obligations to family care and actual
caregiving behavior. Specifically, the first paper showed that filial norms are
malleable and may change over the adult lifespan in response to various intra-
and inter-individual processes. Filial norms peak at midlife and weaken in old
age, especially after the death of both parents, when generational succession
takes place and individuals take the role of the oldest generation in the family
lineage. It further reveals that while filial norms have weakened over historical
time between the 80s and the 90s, the younger generation in the family show
stronger filial norms than their parental generation. The second paper
demonstrated that parents may play a role in determining the amount of care
they may receive from their children in old age when such care is needed.
Parental influence may operate through various mechanisms including strong
emotional bonds, demonstration of the desired caring behavior towards one’s
own parents, and exchange mechanisms involving both provision of financial
support at an earlier point in time and promising bequest in the future.
Intergenerational geographic proximity and frequency of contact are key
mediating factors, facilitating the ability of adult children to provide the required
support. The third paper suggested that filial norms may not always be
accompanied by actual supportive behaviors, even if such help is needed. It
177
demonstrates that about half of the individuals in this study present a discrepant
pattern, whereby they report strong obligations towards eldercare, yet do not
provide high levels of such care to their parents. The remaining subjects divide
about equally into two opposing types of congruent patterns—those who report
weak eldercare obligations and provide low levels of care, and those who are
strongly committed to eldercare and indeed provide high levels of parental
support. Geographic distance and competing demands put adult children at
higher risk of either not reporting high level of commitment or not providing high
levels of parental care even if they are committed to such care. Adult children
who are non-religious are at higher risk of being completely non committed to,
and uninvolved in, parental care.
As a whole, this dissertation demonstrates that intergenerational
relationships, and specifically exchange of support, are conditioned upon the
multiple and complex contexts within which they take place. It further
demonstrates that the normative obligation that may guide such parental
support evolves and changes over the adult lifespan and over historical time.
Combined with the finding that earlier aspects of intergenerational relationships
influence the amount of support adult children are providing to their parents in
old age, these findings demonstrate that intergenerational processes should be
studied longitudinally within a temporal context as well. This dissertation,
therefore, provides evidence to the importance of explaining intergenerational
relationship within a theoretical framework that accounts for such contextual
178
complexity. The proposed conceptual framework --the ecological multi-level
dyadic intergenerational model—provides such a model, as it simultaneously
examines intra-individual change over time within two interacting individuals
(e.g., parent-child) within the micro- through macro-level contexts.
B. Limitations
As described in the second chapter of this dissertation, the dataset
utilized for the analyses–the LSOG--presents several weaknesses that are
important to note. Specifically, because the sample is not nationally
representative and because of the sample design, it under-represents lower-
income families and families from diverse minority ethnic and racial
backgrounds. These weaknesses suggest that more analyses using more
diverse datasets should be performed before these results can be generalized
to the American population as a whole. However, it is important to note that the
intergenerational solidarity model, which represents the basic premise of this
dissertation and was developed based on the LSOG, was further tested on
several other cross-ethnic and cross-country samples. All of these empirical
tests, some of which are still ongoing, reveal that the model is generalizable to
various ethnic and cultural groups. For example, a comparison of the LSOG to
regional samples of Mexican-Americans (Giarrusso, Feng, Silverstein, &
Bengtson, 2001; Silverstein & Chen, 1999) and Welsh from the United Kingdom
(Silverstein, Burholt, Wenger, & Bengtson, 1998) suggested that the
intergenerational model is comparable to these contexts. Other efforts to test
179
this model on diverse populations in the United States involve data collected as
part of the National Study of American Life (NSAL) involving ethnic
comparisons within race. Specifically, Jackson, Antonucci, Sellars, and Brown
(2006) reported on nationally representative data of African American, African
Caribbean, and White populations, and are developing models that test how the
intergenerational solidarity model can be used to study intergenerational
relationships as processes that are historically, culturally, and ethnically
conditioned.
Other efforts to test this model were and are continuously performed
internationally as well. For example, Chen and Silverstein (2000) found the
solidarity model to be applicable to intergenerational relationships in China.
Moreover, The OASIS study, a nationally representative cross national study of
five countries (Norway, England, Germany, U.K., and Israel), utilized the
solidarity model as the basic premise of understanding intergenerational
relationships and repeatedly show more similarities than differences across
countries in the structure and function of intergenerational relationships. Most
of the analyses using these data (e.g., Lowenstein et al., 2004; Lowenstein &
Daatland, 2006) suggested that most differences found are attributed to the
national welfare regimes and the available formal services and specific culture
surrounding the use of these services across the various countries.
180
These cross-ethnic and cross-country empirical examinations suggest that the
LSOG represents an adequate source of examination of the solidarity model
and intergenerational relationships.
In addition, the possible weaknesses of this dataset must be weighed
against the unique advantages it provides, as previously discussed in the
second chapter of this dissertation. Specifically, the study’s longitudinal multi-
generational design and the plethora of variables at the micro- through macro-
levels, allow examination of the research questions posed by this dissertation
as guided by the proposed dyadic multi-level ecological framework.
C. Implications for Current Knowledge and Future Research
This dissertation adds to the cumulative body of knowledge and bears
implications for future research by demonstrating that the relationship between
filial norms and filial behavior is a complex relationship and is conditioned upon
various contextual and temporal factors. Social researchers and family
scholars who plan to use these concepts, therefore, should attempt to consider
these contextual and temporal factors. Specifically, the dissertation as a whole
collectively makes the following contributions:
First, the dissertation asserts that filial norms are not trait-like qualities.
Rather, as demonstrated by the first paper, filial commitment to parental care
ebbs and flows through a process of intra-individual change over the adult
lifespan, as well as in response to external circumstances—specifically, for
example, previous caregiving experience and generational succession. This is
181
a significant contribution as this dissertation presents the first analysis that
examined change over the lifespan and over historical time in filial norms, as
well as variability in such elder commitment across successive generations.
Previous analyses treated filial norms as a constant predictor of filial behavior
and assumed that it is invariable over the lifespan, across cohorts, and over
historical time. The assertion that such norms are malleable is further
supported by the second paper’s finding that parents’ level of commitment in an
earlier point in time does not predict the same level of commitment on the part
of the children at a later point in time. In other words, the similarity in strength
with which parents and children support the responsibility of adult children to
provide parental care may not be a good indication of a successful socialization
of caring behavior. Understanding that filial norms are dynamic and malleable
has theoretical, practical, and methodological implications. Theoretically, this
finding shows support to developmental theories, suggesting that indeed filial
norms peak at midlife. Because adult children are most likely to be called upon
to provide parental support at midlife, this finding may bear practical
implications, suggesting that adult children may indeed be ready to fulfill their
filial role in times of need. Similar to this study, future research addressing filial
norms should use research methodologies that allow modeling this construct as
a dynamic, rather than static, factor.
Second, this dissertation demonstrates that filial norms may not be a
direct predictor of actual supportive behavior. The relationship between filial
182
norms and filial behavior is complex and is conditioned on several
circumstantial and contextual factors. As demonstrated by the third paper, filial
norms may not necessarily be enacted into actual supportive behavior. In fact,
when testing the data in the third paper, for patterns of congruence and
discrepancy between filial norms and filial behavior, half of the subjects fell into
the dissonant category—a category that reflects strong filial commitment that is
not accompanied by actual supportive behavior. Furthermore, the second
paper suggested that when other aspects in the intergenerational relationships
are considered, filial norms do not play a significant role in predicting supportive
behavior. Specifically, this study shows that affective bonds between children
and parents over time are a powerful predictor of both instrumental and
emotional support, regardless of filial commitment. But perhaps even more
interesting in supporting this assertion is the finding that parental care can be
modeled and encouraged through the mechanism social learning. The second
study shows that adult children are likely to demonstrate caring behavior
towards their parents after witnessing such behavior from their own parents to
their grandparents. Simply put, this dissertation proposes that in order for
parents to affect their children to provide support to them, it is not enough to
“talk the talk”, by socializing for caring behavior, but rather parents need to
actually “walk the walk”, by modeling the desired support in their own behavior.
At the same time, however, this dissertation still provides some evidence
that filial commitment may still be an important factor to consider when
183
predicting support. Based on the third paper, non-religiosity may put one at
higher risk of being uninvolved in parental care by showing weak filial
obligations and low levels of parental support. As discussed in detail in the third
paper, religiosity may promote caring behavior in general, and parental care
specifically, in more than one way. It may be that religious involvement
provides more opportunity to participate in such caring behavior, in which case,
religiosity operates as a social learning mechanism. But it is also possible that
people develop stronger filial norms because they are socialized through
various religious beliefs that one should care for others in general, and for one’s
parents specifically. While it is impossible to tease out which mechanism works
based on this specific study, it is possible that religion provides a socialization
tool, by which adult children follow certain moral or religious convictions, which
in turn, strengthen their filial commitment and lead to higher levels of parental
support. Future studies may attempt to further decipher this interesting
relationship between religiosity, filial norms and filial behavior.
Other important factors that should be considered involve various
circumstantial factors of both the adult children and their aging parents. As
suggested by the dyadic ecological model, I indeed found that competing
demands on the part of the adult children in the form of having children (paper
three) as well as marriage (paper two) may decrease the likelihood of
supportive behavior, whereas the level of health needs and widowhood (paper
two) on the part of the parents may increase levels of support. This
184
dissertation, therefore, makes it clear that the relationship between filial norms
and filial behavior is more complex than once thought and that it should not be
analyzed out of context and without considering other important factors such as
the quality of the relationship between the children and their parents over time,
religious practices, and family and generation membership, as well as specific
circumstantial factors.
Third, the dissertation provides evidence to suggest that opportunity in
the form of frequency of contact across generations and geographic proximity
between adult children and their parents play a key role in predicting actual
supportive behavior. Each opportunity factor has a unique contribution to actual
supportive behavior. As demonstrated by the second paper, geographic
proximity is a strong mediator between motivations for parental care and actual
provision of instrumental support. One is more likely to provide instrumental
support, in the form of help with household chores and transportation and help
when a parent is sick, if one lives close to that parent. This assertion is further
supported by findings from the third paper, which strongly suggests that
geographic distance is the most powerful factor in putting an adult child at risk
of not developing filial norms or not acting upon those norms. Frequency of
contact, on the other hand, and mostly in the form of phone conversations, is
found in the second paper to be a mediating factor between various motivations
for supportive behavior and emotional support provided to mothers. In other
words, mothers may still enjoy emotional support and receive information and
185
advice from their children even if they live farther away, as long as they
communicate frequently. Understanding the significant, yet differing, roles of
frequency of contact and geographic distance on provision of parental support
may have important policy and practice implications. Increasing migration
patterns (Crimmins & Ingegneri, 1990) suggest that aging parents and adult
children are less likely to live in close geographic proximity to one another. At
the same time, however, communications forms, such as phone and e-mail, are
continuously improving and becoming more accessible. While families are
viewed by policymakers as social units that are ready, willing, and able to meet
the growing needs of the elderly population (Parrott & Bengtson, 1999), this
dissertation suggests that families may be more available to meet the emotional
needs of their elders and perhaps less available to meet more instrumental
needs. Such knowledge may inform policymakers as they formulate and
amend new social and aging policies that guide the development and
implementation of formal services. This dissertation suggests that while
families may still provide elders with information and advice and emotional
support, a greater effort should be directed towards developing formal services
that provide more instrumental types of support such as transportation and help
with household chores.
Future studies should attempt to include the various contexts within
which filial norms are developed and are enacted to actual care, including
considering the temporal context by continuing to conduct longitudinal studies.
186
Filial norms should be addressed as a dynamic factor which changes over the
lifespan and in response to inra- and inter-individual processes. Because the
dataset used for this analysis under-represents minority groups and individuals
from diverse ethnic and racial backgrounds, further analyses should be
conducted before one can state that the results are universal to individuals in
the Unites States. It would further be beneficial to conduct similar analyses
using cross-national dataset and attempt to explain the role of national context
and especially welfare regime in shaping the development of filial norms and
their enactment into actual support.
187
Bibliography
Aboderin, I. (2004). “Conditionality” and “limits” of filial obligation: Conceptual
levers for developing a better understanding of the motivational bias and
societal shifts or patterns in old age family support. Oxford Institute of
Ageing Working Paper No. 604, Institute of Ageing, Oxford
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 the Family, 62(4), 911-926.
Alwin, D. F. (1990). Cohort replacement and changes in parental socialization
values. Journal of Marriage and the Family, 52, 347-360.
Alwin, D. F. (1998). The political impact of the Baby Boom: Are there persistent
generational differences in political beliefs and behaviors? Generations,
22(1), 46-54.
Alwin D. F., & McCammon, R. J. (2003). Generations, cohorts, and social
change. In J. T. Mortimer & M. J. Shanahan (Eds.), Handbook of the life
course (pp. 23-50). New York: Kluwer Academics/Plenum.
Antonucci, T. (1990). Social supports and social relationships. In R. H. Binstock
&. L. K. George (Eds.), Handbook of aging and the social sciences (Third
ed., pp. 205-226). New York: Academic Press.
Bellah, R. N., Madsen, R., Sullivan, W. M., Swidler, A., & Tipton, S. M. (1985).
Habits of the heart: Individualism and commitment in American life.
Berkeley, CA: University of California Press.
Bengtson, V. L. (1975). Generational and family effects in value socialization.
American Sociological Review, 40, 358-371.
Bengtson, V.L. (Ed.) (1996). Adulthood and aging: Research on continuities
and discontinuities. New York, NY: Springer Publishing Company.
Bengtson, V.L. (2001). Beyond the nuclear family: The increasing importance
of multigenerational bonds. Journal of Marriage and Family, 63(1), 1-16.
Bengtson, V. L., Acock, A. L., Allen, K. R., Dilworth-Anderson, P., & Klein, D. M.
(2005). Theory and theorizing in family research: Puzzle building and
puzzle solving. In V.L. Bengtson, A. Acock, K.Allen, P. Dilworth-Anderson,
188
& D. Klein (Eds.) Sourcebook of Family Theory and Research (pp. 3-33).
Thousand Oaks, CA: Sage Publications.
Bengtson, V. L., & Allen, K. (1993). The life course perspective applied to
families over time. In P. Boss, W. Doherty, R. LaRossa, W. Schumm, & S.
Steinmetz (Eds.), Sourcebook of family theories and methods: A contextual
approach (pp. 469-498). New York: Plenum Press.
Bengtson, V.L., Biblarz, T.J., & Roberts, E.L. (2002). How families still matter.
Cambridge, UK; Cambridge University Press.
Bengtson, V.L., & Mangen, D.J. (1988). Family intergenerational solidarity
revisited: Suggestions for future managements. In D.J., Mangen, V.L.
Bengtson, & P.H. Landry (Eds.). Measurement of intergenerational relations
(pp. 222-238). Beverly Hills, CA: Sage Publications.
Bengtson, V.L., & Roberts, R.E.L. (1991). Intergenerational solidarity in aging
families: An example of formal theory construction. Journal of Marriage and
the Family, 53, 856-870.
Blenkner, M. (1965). Social work and family relationship in late life. In E.
Shanas & G. Streib (Eds.), Social structure and the family (pp. 46-61).
Englewood Cliffs, NJ: Prentice – Hall.
Brody, E.M. (1981). Women in the middle and family help to older people. The
Gerontologist, 21, 471-480.
Brody E.M., & Brody, S.J. (1989). The informal system of health care. In F.
Eisdorfer, D.A. Kessler, & A.N. Spector (Eds.) Caring for the elderly:
Reshaping health policy (pp. 259-277). Baltimore, MD: John Hopkins
University Press.
Bromely, M.C., & Blieszner, R. (1997). Planning for long term care: Filial
behavior and relationship quality of adult children with independent parents.
Family Relations, 46(2), 155-162.
Bronfenbrenner, U. (1979). The ecology of human development. Cambridge,
MA: Harvard University Press.
Bronfenbrenner, U. (1986). Recent advances in research on the ecology of
human development. In R. K. Silberman, K. Eyferth, & G. Rudinger (Eds.).
Developmental action in context: Problem behavior and normal youth
development (pp. 287-309). Heidelberg & New York: Springer-Verlag.
189
Bronfenbrenner, U. (1997). Ecology of the family as a context for human
development: Research perspective. In J.L. Paul, M.Churton, H. Rosselli-
Kostoryz, W.C. Morse, K. Mafro, C. Lavely, & D. Thoms (Eds.), Foundations
of special education (pp. 49-83). Pacific Grove, CA: Brooks/Cole
Publishing Company.
Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models:
Applications and data analysis methods. Thousand Oaks, CA: Sage.
Byrne, B.M. (2001). Structural equation modeling with AMOS: Basic concepts,
applications, and programming. Mahwah, NJ: Lawrence Erlbaum
Associates.
Bubolz, M., & Sontag, S. (1993). Human ecology theory. In P.G. Boss, W.J.
Doherty, R.LaRossa, W.B. Scham, & S.K.Steinmetz (Eds.), Sourcebook of
families, theories, and methods: A contextual approach (pp. 419-448). New
York: Plenum.
Burr, J.A, & Mutchler, J.E. (1999). Race and ethnic variation in norms of filial
responsibility among older persons. Journal of Marriage and the Family,
61(3), 674-687.
Call, V.R.A., & Heaton, T.B. (1997). Religious influence on marital stability.
Journal of the Scientific Study of Religion, 36, 382-392.
Cantor, M. H. (1983). Strain among caregivers: a study of experience in the
United States. Gerontologist, 23(6), 597-604.
Caputo, R. K. (2005). Inheritance and intergenerational transmission of
parental care. Marriage and Family Review, 37 (1/2), 107-127.
Casper, L. M., & Bianchi, S. M. (2002). Continuity and change in the American
family. Thousand Oaks, CA: Sage.
Cherlin, A., & Frustenberg, F. (1986). The new American grandparent: A place
in the family, a life apart. New York: Basic Books.
Chen, X., & Silverstein, M. (2000). Intergenerational social support and the
psychological well-being of older parents in China. Research on Aging, 22,
43-66.
190
Cicirelli, VG. (1981). Helping elderly people: Role of adult children. Boston,
MA: Alburn House.
Cicirelli, V. G. (1983). A comparison of helping behavior to elderly parents of
adult children with intact and disrupted marriages. The Gerontologist, 23,
619-625.
Cicirelli, V.G. (1988) A measure of filial anxiety regarding anticipated care of
elderly parents. The Gerontologist, 23(4), 478-482.
Cicirelli, V.G. (1990). Family support in relation to health problems of the
elderly. In T.H. Brubaker. (Ed.) Family relationships in later life. (pp. 212-
228). Newbury Park, CA: Sage Publications.
Clogg, C.C. (1995). Latent class models. In G. Arminger, C.C. Clogg, & M.E.
Sobel (Eds.), Handbook of statistical modeling for the social and behavioral
science (pp. 311-360). New York: Plenum Press.
Clogg, C.C., & Goodman, L.A. (1984). Latent structure analysis of a set of
multidimensional contingency tables. Journal of American Statistical
Association, 79, 762-771.
Coleman, J.S. (1990). Foundations of social theory. Cambridge, MA: Belkhap
Press of Harvard University Press.
Connidis, I. A. (2001). Family ties and aging. Thousand Oaks, CA: Sage
Publications.
Crimmins, E. (2004). Trends in the health of the elderly. Annual Reviews of
Public Health, 25, 79-98.
Crimmins, E., & Ingengneri, D. (1990). Interaction and living arrangements of
older parents and their children: Past trends, present determinants, future
implications. Research on Aging, 12, 3-35.
Crosnoe, R., Mistry, R.S., & Elder, G.H., Jr. (2002). Economic disadvantages,
family dynamics, and adolescent enrollment in higher education. Journal of
Marriage Family, 64, 690 - 702.
Copen, C., Biblarz, T.J., Silverstein, M., & Bengtson V.L. (2005, August). The
ties that bind: Intergenerational transmission of religious values within
American families. American Sociological Association Annual Meeting,
Philadelphia, PA.
191
Daatland, S.O., & Herlofsson, K. (2003). ‘Lost solidarity’ or ‘changed solidarity’:
A comparative European view of normative family solidarity. Ageing and
Society, 23(5), 537-561.
Davey, A., Janks, M., & Salva, J (2004). Antecedents of intergenerational
support: Families in context and families as context. In M. Silverstein (Ed.),
Annual Review of Gerontology and Geriatrics, Volume 24: Intergenerational
Relations Across Time and Place (pp. 29-54). New York: Springer
Publishing Company.
Easterlin, R., Schaeffer, C., & Macunovich, D. (1993). Will the baby boomers
be less well-off than their parents? Income, wealth, and family
circumstances over the life cycle. Population and Development Review,
19(3), 497-522.
Eggebeen, D.J. (1992). Family structure and intergenerational exchange.
Research on Aging, 14(4), 427-447.
Eggebeen, D.J., & Davey, A. (1998). Do safety nets work? The role of
anticipated help in times of need. Journal of Marriage and the Family, 60(4),
939-950.
Elder, G. H., Jr. (1992). Models of the life course. Contemporary Sociology: A
Journal of Reviews, 21(5), 632-635.
Elder, G. H., Jr. (1994). Time, human aging, and social change: Perspectives
on the life course. Social Psychology Quarterly, 57, 4-15.
Elder, G. H., Jr., & Johnson, M. K. (2002). Perspectives on human
development in context. In C. von Hofsten, & L. Backman (Eds),.
Psychology at the turn of the millennium, vol. 2: Social, developmental, and
clinical perspectives (pp. 153- 172). Florence, KY: Taylor &
Frances/Routledge.
Finch, J., & Mason, J. (1991). Obligations of kinship in contemporary Britain: Is
there Normative agreement? The British Journal of Sociology, 42, 345-367.
Finley, N.J., Roberts, M.D., & Banahan, B.F., III (1988). Motivations and
inhibitors of attitudes of filial obligations towards aging parents. The
Gerontologist, 28 (1), 73-78.
192
Furstenberg, F. F., & Kaplan, S. B. Social capital and the family. (2004). In J. L.
Scott, J. K. Treas, & M. P. Richards (Eds.). The Blackwell companion to the
sociology of families. (pp. 218-232). Malden, Mass.: Blackwell Publishers.
Ganong, L. H., & Coleman, M. (1999). Changing families, changing
responsibilities: Family obligations following divorce and remarriage.
Mahwah, NJ: Erlbaum.
Ganong, L., Coleman, M., McDaniel, A.K., & Killian, T. (1998). Attitudes
regarding obligations to assist older parent or stepparents following later-life
marriages. Journal of Marriage and the Family, 60(3), 595-610.
Gans, D., & Schiamberg, L. B. (1997, August). An ecological framework for
risk factors of elder abuse and family caregiving. Poster session presented
at the World Congress of Gerontology, Adelaide, Australia.
George, L. K., Larson, D. B., Koenig, H. G., McCullough, M. (2000). Spirituality
and health: State of the evidence. Journal of Social and Clinical
Psychology, 19, 102-116.
Giarrusso, R., Feng, D., Silvertein, M., & Bengtson, V.L. (2001). Grandparent-
adult grandchild affection and consensus. Journal of Family Issues, 22,
456-478.
Giarrusso, R., Silverstein, M., Gans. D., & Bengtson, V.L. (2005). Aging
Parents and Adult Children: New Perspectives on Intergenerational
Relationships. In M. Johnson, V.L. Bengtson, P. Coleman, & T. Kirkwood.
(Eds.). The Cambridge Handbook on Age and Ageing. London, England:
Cambridge Press.
Glass, J., Bengston, V.L., & Dunham, C.C. (19860. Attitude similarity in three-
generation families: socialization, status inheritance, or reciprocal influence?
American Sociological Review, 51, 685-698.
Guberman, N. (2003, November). Beyond demography: The norms and
values of family solidarity in theory and in practice. Paper presented at the
56
th
Annual Meeting of the Gerontological Society of America, San Diego,
California.
Hamon, R.R. (1992). Filial role enactment by adult children. Family Relations,
41(1), 91-96.
193
Hareven, T. K., (1995). Historical perspectives on the family and aging. In R.
Blieszner & V.H. Bedford (Eds.), Handbook of aging and the family (pp. 13-
31). Westport, CT: Greenwood Press.
Hareven, T. K. (1996). Introduction: aging and generational relations over the
life course. In T.K. Hareven (Ed.), Aging and generational relations over
the life course (pp 1-12). New York: Walter de Gruyter.
Heaton, T.M., & Pratt, E.L. (1990). The effects of religious homogamy on
marital satisfaction and stability. Journal of Family Issues, 11, 191-207.
Heckman, J. J. (1976). The common structure of statistical models of
truncation, sample selection and limited dependent variables and a simple
estimator for such models. Annals of Economic and Social Measurement, 5,
475-492.
Himes, C. (1994). Parental caregiving by adult children: A demographic
perspective. Research on Aging, 16(2), 191-211.
Hirdes, J.P., & Strain, L.A. (1995). The balance of exchange in instrumental
support with network members outside the household. Journal of
Gerontology: Social Sciences, 50B, S134-S142.
Hoge, D.R., Petrillo, G.H., & Smith, E.I. (1982). Transmission of religious and
social values from parents to teenage children. Journal of Marriage and the
Family, 44(3), 569-580.
Hogan, D.P., Eggebeen, D.J., & Clogg, C.C. (1993). The structure of
intergenerational exchanges in American families. The American Journal of
Sociology, 98(6), 1428-1458.
Horn, J. L., & McArdle, J. J. (1980). Perspectives on mathematical/statistical
model building (MASMOB) in research on aging. In L. W. Poon (Ed.), Aging
in the 1980s (pp. 503-541). Washington, DC: American Psychological
Association.
Horowitz, A. (1985). Sons and daughters as caregivers to older parents:
Differences in role performance and consequences. The Gerontologist, 25,
612-617.
Ikkink, K.K., van Tilburg, T., & Knipscheer, C.P.M. (1999). Perceived
instrumental support exchanges in relationships between elderly parents
194
and their adult children: Normative and structural explanations. Journal of
Marriage and the Family, 61, 831-844.
Jackson, J.S., Antonucci, T.C., Sellars, B., & Brown, E. (November, 2006).
Population aging, ethnicity, public policy and family support in the United
States. Paper presented at the Annual Gerontological Society of America,
Dallas: Texas.
Kain, E.L. (1990). The myth of family decline. Lexington, M.A: D.C Heath.
Kunemund, H., & Rein,M. (1999). There is more to receiving than needing:
Theoretical arguments and empirical explorations of crowding in and
crowding out. Ageing and Society, 19, 93-121.
Koenig, H.G., McCullough, M.E., & Larson, D.B. (2001). Handbook of religion
and health. New York: Oxford University Press.
Krosnick, J. A., & Alwin, D. F. (1989). Aging and susceptibility to attitude
change. Journal of Personality and Social Psychology, 57, 416-425.
Laditka, J.N., & Laditka, S.B. (2001). Adult children helping older parents:
Variation in the likelihood and hours by gender, race and family role.
Research on Aging, 23(4), 429-456.
Lee, G.R., Netzer, J.K., & Coward, R.T. (1994). Filial responsibility
expectations and patterns of intergenerational assistance. Journal of
Marriage and the family, 56(3), 559-565.
Lee, G. R., Peek, C. W., & Coward, R. T. (1998). Race differences in filial
responsibility expectations among older parents. Journal of Marriage and
the Family, 60, 404-412.
Litwak, E. Silverstein, M., Bengtson, V., & Wilson-Hirst, Y. (2003). Theories
About families, organizations, and social support (pp. 27-35), In V. L.
Bengtson, & Lowenstein, A. (Eds.), Global aging and challenges to families,
New York: Aldine de Gruyter.
Logan, J. R., & Spitze, G. D. (1995). Self-interest and altruism in
intergenerational relations. Demography, 32, 353-364.
Loomis, L.S., & Booth, A. (1995). Multigenerational caregiving and well-being:
The myth of the Sandwich generation. Journal of Family Issues, 16(2), 131-
148.
195
Lowenstein, A., & Daatland S.O. (2006). Filial norms and family support in a
comparative cross-national context: Evidence from the OASIS study.
Ageing and Society, 26, 203-233.
Lowenstein, A. Katz, R., & Daatland, S.O. (2004). Filial norms and
intergenerational support in European and Israeli comparative perspective.
In M. Silverstein (Ed.), Annual Review of Gerontology and Geriatrics,
Volume 24: Intergenerational Relations Across Time and Place (pp. 200-
223). New York: Springer Publishing Company.
Mannheim, K. (1928/1952). The problem of generations [1928]. In D.
Kecskemeti (Ed.), Essays on the sociology of knowledge (pp. 276-322).
London: Routledge & Kegan Paul.
Maruyama, G. (1998). Basics of structural equation modeling. Newbury Park:
Sage Publications.
McArdle, J. J., Hamagami, F., Jones, K., Jolesz, F., Kikinis, R., Spiro, A., &
Albert, M. S. (2004). Structural modeling of dynamic changes in memory
and brain structure using longitudinal data from the Normative Aging Study.
Journal of Gerontology: Psychological Sciences, 59, P294-P304.
Miller, A.S., & Stark, R. (2002). Gender and religiousness: Can socialization
explanations be saved? American Journal of Sociology, 107, 1399-1423.
Miller, R. B., & Glass, J. (1989). Parent-child attitude similarity across the life
course. Journal of Marriage and the Family, 51, 991-997.
Motel-Klingebiel, A., Tesch-Roemer, C., & von Kondratowitz, H.Y. (2005).
Welfare states do not crowd out the family: evidence for mixed responsibility
from comparative analyses. Ageing and Society, 25, 863-882.
Motel-Klingebiel, A., Tesch-Roemer, C., & von Kondratowitz, H.Y. (2003). The
role of family for quality of life in old age: A comparative perspective (pp.
327-354), In V. L. Bengtson, & Lowenstein, A. (Eds.), Global Aging and
Challenges to Families. New York: Aldine de Gruyter.
Morgan, D.L., Schuster, T.L., & Butler, E.W. (1991). Role reversals in the
exchange of social support. Journals of Gerontology, 46(5), S278-S287.
196
Murray, P.D., Lowe, J.D., & Horne, H.L. (1995). Assessing filial maturity
through the use of the filial anxiety scale. The Journal of Psychology, 129
(5), 519-529.
Mutschler, P.H., (1994). From executive suit to production line: How employees
in different occupations manage elder care responsibilities. Research on
Aging, 16(1), 7-26.
Myers, S. M. (2004). Religion and intergenerational assistance: Distinct
differences by adult children’s gender and parent’s marital status.
Sociological Quarterly, 45 (1), 67-89.
Nydegger, C.N. (1991). The development of parental and filial maturity. In K.
Pillemer & K. McCarthney (Eds.), Parent-child relations throughout life (pp.
93-112). Hillside, NJ: LEA Publishers
.
Parrott, T.M., & Bengtson, V.L. (1999) The effects of earlier intergenerational
affection, normative expectations, and family conflict on contemporary
exchanges of help and support. Research on Aging, 21(1), 73-105.
Pearce, L.D., & Axinn, W.G. (1998). The impact of family religious life on the
quality of mother-child relations. American Sociological Review, 63, 810-
828.
Peek, M.K., Coward, R.T., Peek, C.W., & Lee, G.R. (1998). Are expectations
for care related to the receipt of care? An analysis of parent care among
disabled elders. Journal of Gerontology: Social Sciences, 53B(3), S127-
S136.
Popenoe, D. (1988). Disturbing the nest: Family change and decline in modern
societies. New York: Aldine De Gruyter.
Putnam, R. D. (1995). Bowling alone: America’s declining social capital.
Journal of Democracy, 6(1), 65-71.
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models:
Applications and data analysis methods (2
nd
ed.). Thousand Oaks, CA:
Sage.
Raftery, A.E. (1986). Choosing models for cross-classifications. American
Sociological Review, 51, 145-146.
197
Ribar, D.C., & Wilhelm, M.O. (2006). Exchange, role modeling and the
intergenerational transmission of elder support attitudes: Evidence from
three generations of Mexican-Americans. Journal of Socio-Economics,
35(3), 514-531.
Riley, M.W. (1973). Aging and cohort succession: Interpretations and
misinterpretations. Public Opinion Quarterly, 37, 843-861.
Riley, M.W. (1983). The family in an aging society: A matrix of latent
relationships. Journal of Family Issues, 4, 439-454.
Roberts, R. E. L., & Bengtson, V. L. (1999). The social psychology of values:
Effects of individual development, social change, and family transmission
over the life span. In C. Ryff & V. Marshall (Eds.), The self and social
processes in aging (pp. 453-482). New York: Springer.
Rosenthal, C. J. (2004). Help to older parents and parents-in-law: Does paid
employment constrain women's helping behaviour? Canadian Journal on
Aging, S97-S112
Rossi, A. S., & Rossi, P. H. (1990). Of human bonding: Parent-children
relationship across the life course. New York: Aldine de Gruyter.
Ryder, N. B. (1965). The cohort as a concept in the study of social change.
American Sociological Review, 30, 843-861.
Sarkisian, N., & Gerstel, N (2004). Kin support among Black and White: Race
and family organization. American Sociological Review, 69 (6), 812-837.
Schiamberg L. B., & Gans, D. (2000). Elder abuse by adult children: An
ecological framework for understanding contextual risk factors and the
intergenerational character of quality of life. The International Journal of
Aging and Human Development, 50(4), 329-359.
Schiamberg, L. & Gans, D. (2005). An ecological perspective to an
intergenerational family problem. In V.L. Bengtson, A. Acock, K.Allen, P.
Dilworth-Anderson, & D. Klein (Eds.) Sourcebook of Family Theory and
Research (pp. 332-334). Thousand Oaks, CA: Sage Publications.
Sears, D. O. (1981). Life-stage effects on attitude change, especially among
the elderly. In J. M. March, S. B. Kiesler, J. M. Morgan, & V. K.
Oppenheimer (Eds.), Aging (pp. 183-204). New York: Academic Press.
198
Shanas, E. (1979). The Robert W. Kleemeier Award lecture. Social myth as
hypothesis: The case of the family relations of old people. Gerontologist,
19(1), 3-19.
Shuey, K. & Hardy, M.A. (2003). Assistance to aging parents and parents-in-
law: Does lineage affect family allocation decisions? Journal of Marriage
and Family, 65 (2), pp. 418-431.
199
Siegel, J. (1993) Basic Demographic Dimensions of an Aging Population, a
Generation of Change. New York, NY: Russell Sage.
Silverstein, M. (2005). Testing theories about intergenerational exchange. In
V.L. Bengtson, A. Acock, K. Allen, P. Dilworth-Anderson, & D. Klein (Eds.)
Sourcebook of family theory and research (pp. 407-413). Thousand Oaks,
CA: Sage Publications.
Silverstein, M., & Bengtson, V.L. (1997). Intergenerational solidarity and the
structure of adult child-parent relationships in American families. The
American Journal of Sociology, 103, 429-460.
Silverstein, M., Burholt, V, Wenger, G.C., & Bengtson, V.L. (1998). Parent-
child relations among very old parents in Wales and the United States: A
test of modernization theory. Journal of Aging Studies, 12, 387-410.
Silverstein, M., & Chen, X. (1999). The impact of acculturation in Mexican
American families on the quality of adult grandchild-grandparent
relationships. Journal of Marriage and the Family, 61, 188-199.
Silverstein, M., Conroy, S., Wang, H., Giarrusso, R, & Bengtson, V. L. (2002).
Reciprocity in parent-child relations over the adult life course. Journal of
Gerontology: Social Sciences: 57, S3-S13.a
l
Silverstein, M., Gans, D., & Yang, F.M. (2006) Filial Support to Aging Parents:
The Role of Norms and Needs. Journal of Family Issues, 27, 1068-1084.
Silverstein, M., & Litwak, E. (1993). A task specific typology of
intergenerational family structure in later life. The Gerontologist, 33(2), 258-
264.
Silverstein, M. & Parrott, T.M. (2001). Attitudes toward government policies
that assist informal caregivers: The link between personal troubles and
public issues. Research on Aging, 23(3), 349-375.
Silverstein, M., Parrott, T.M., & Bengtson, V.L. (1995). Factors that predispose
middle aged sons and daughters to provide social support to older persons.
Journal of Marriage and the Family, 57, 465-475.
Silverstein, M., & Waite, L. (1993). Are Blacks more likely than White to
receive and provide social support in middle and old age? Yes, no, and
maybe so. Journals of Gerontology: Social Sciences, 48, S212-S222.
200
Smith, C. (2005). Soul searching: The religious and spiritual lives of American
teenagers. New York: Oxford University Press.
Sorenson, S., & Zarit, S.H. (1996). Preparation for caregiving:: A study of
multigenerational families. International Journal of Aging and Human
Development, 42(1), 43-63.
Spitze, G., & Logan, J. (1989). Gender differences in family support: Is there a
payoff? Gerontologist, 29, 108-113.
Spitze, G., & Logan, J. (1991). Employment and filial relations: Is there a
conflict? Sociological Forum, 6(4), 681-697.
Stein, C.H., Wemmerus, V.A., Ward, M., Gaines, M.E., Freeberg, A.L., & Jewell,
T. (1998). “Because they’re my parents”: An Intergenerational Study of felt
obligation and parental caregiving. Journal of marriage and the Family, 60,
611-622.
Stoller, E.P. (1983). Parental caregiving by adult children. Journal of Marriage
and the Family, 45, 851-858.
Stone, R., Cafferta, G.L., Lee G., & Sangl, J. (1987). Caregivers of the frail
elderly: A National profile. The Gerontologist, 27 (5), 616-626.
Szinovacz, M. (1997). Adult children taking parents into their homes: Effects of
childhood living arrangements. Journal of Marriage and the Family, 59, 700-
717.
Taylor, R.J., Chatters, L.M., & Jackson, J.S. (1993). A profile of familial
relations among 3-generation Black families. Family Relations, 42, 332-341.
Thornton, A., Axinn, W.G., & Hill, D.H. (1992). Reciprocal effects of religiosity,
cohabitation, and marriage. American Journal of Sociology, 98, 628-651.
Thornton, A., & Young-DeMarco, L. (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.
Troll, L.E. (Ed). (1986). Family issues in current gerontology. New York, NY:
Singer.
201
Uhlenberg, P.R., & Cooney, T.M. (1990). Family size and mother-child
relations in later life. Gerontologist, 30, 618-625.
Vermunt, J.K., & Magidson, J. (2000). Latent Gold user guide. Belmont, Mass:
Statistical Innovations Inc.
Walker, A.J., Pratt, C.C., Shinn, H., & Jones, L.L. (1990). Motives for parental
caregiving and relationship quality. Family Relations, 39, 51-56.
Whitbeck, L.B, & Gecas, V. (1988). Value attributes and value transmission
between parents and children. Journal of Marriage and the Family, 50(3),
829-840.
Whitbeck, L., Hoyt, D.R., & Huck, S.M. (1994). Early family relationships,
intergenerational solidarity, and support provided to parents by their adult
children. Journal of Gerontology: Social Sciences, 49(2), 585-594.
Wilcox, B.W. (1998). Conservative Protestant childbearing: Authoritarian or
authoritative? American Sociological Review, 63, 796-809.
Wolf, D.A., Freedman, V., & Soldo, B.J. (1997). The division of family labor:
Care for elderly parents. The Journals of Gerontology, Social Sciences,
52B, 102-109.
Wothke, W. (2000). Longitudinal and multigroup modeling with missing data. In
T. D. Little, K. U. Schnabel, & J. Baumert (Eds.), Modeling longitudinal and
multilevel data: Practical issues, applied approaches, and specific examples
(pp. 219-240). Mahwah, NJ: Lawrence Erlbaum Associates, Publishers.
Zarit, S.H., & Reid J.D. (1994). Family caregiving and the older family. In C.B.
Fischer & R.M. Lerner (Eds.) Applied Developmental Psychology (pp. 237-
267). New York, NY: MacGraw-Hill.
Abstract (if available)
Abstract
This dissertation aimed to explain how the various contexts of family and generation membership, religious affiliation and religiosity, and historical period, shape the degree to which adult children feel obligated to, and actually support, their aging parents. More specifically, the objectives were: (1) to examine change across the lifespan in the strength with which individuals feel adult children should be committed to parental care
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Filial expectations and social exchange patterns among older Taiwanese parents and their adult children
PDF
The measurement, life course patterns, and outcomes of intergenerational ambivalence among parent-adult child dyads
PDF
Falls among older adults: characteristics of fallers, predictors of falls, and the impact of falls on health care and long-term care utilization
PDF
Intergenerational social support and the psychological well-being of older parents in China
PDF
Transmission and transitions: young adults' beliefs, values and life course transitions in familial context
PDF
Children 's migration and the financial, social, and psychological well-being of older adults in rural China
PDF
The immediate and long term legacy of relationships with grandparents for the well-being of grandchildren
PDF
The effect of body weight on mortality: different countries and age groups
PDF
Intergenerational transmission of values and behaviors over the family life course
PDF
Childlessness and psychological well-being across life course as manifested in significant life events
PDF
Intergenerational coresidence of older adults in contemporary Japan: traditional cultural norms in divergent styles
PDF
Internet communication use, psychological functioning and social connectedness at older ages
PDF
Translating research into practice: a community-based medication management intervention
PDF
Longitudinal change in active life expectancy: the longitudinal studies of aging 1984-2000
PDF
Racial/ethnic variation in care preferences and care outcomes among United States hospice enrollees
PDF
Health differences between the elderly in Japan and the United States
PDF
Palliative care: breaking the knowledge barrier among community-based older adults
PDF
International sex and age differences in physical function and disability
PDF
Marital quality, gender, and biomarkers of disease risk in the MIDUS cohort
PDF
Residential care in Los Angeles: policy and planning for an aging population
Asset Metadata
Creator
Gans, Daphna
(author)
Core Title
Normative obligations and parental care in social context
School
Leonard Davis School of Gerontology
Degree
Doctor of Philosophy
Degree Program
Gerontology
Publication Date
04/05/2007
Defense Date
02/16/2007
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
filial norms,intergenerational relations,OAI-PMH Harvest
Language
English
Advisor
Silverstein, Merril (
committee chair
), Bengtson, Vern (
committee member
), Biblarz, Timothy J. (
committee member
), Crimmins, Eileen M. (
committee member
)
Creator Email
dgans@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-m352
Unique identifier
UC1199934
Identifier
etd-Gans-20070405 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-399343 (legacy record id),usctheses-m352 (legacy record id)
Legacy Identifier
etd-Gans-20070405.pdf
Dmrecord
399343
Document Type
Dissertation
Rights
Gans, Daphna
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
Tags
filial norms
intergenerational relations