'Z‘J‘Jflbtfih'f: -? . fl. fig; ‘1
M J 3‘. " 7 ' 1 1; “-1 ,o - * : .,‘ {0‘
‘1 92:53.43. .fb 2“." , k “ah I 7 ‘ . ‘4 ‘1. 6:4

T ‘ 10;?
‘9‘;

A.

. . 1- 9:2
< 'v: \. 1 -:
fivxwxww.‘ ..r._§'
1 mwxwwwm

1- «'2 (11‘ .1. f,
"I. “a; f .
*1 9‘3

'3» 1 .

£-’L?"fi;’av~‘?:i;:j}.§5/L;c§ y '9

; 3" {gm-i: tea-mi: .. J“ . “£3“

.‘1 \ >~ v :59: .w. I c. . 17.34;:

ugfing-fiy « fl ~‘ KW“ I .Qfiti‘agka, :35; a»

a %" $ch ,1 '1- gncc‘; -. 9.21%
I? u». V.» 1.3.5 '4

, ~ % . 3., I 7 - ‘ I . ,
€16; . . _ . f: . . ‘13:: "9 ‘ ’1
$975: ' ‘ ' "7 1 7 ‘ 7 '.

3- 3 V‘ ‘ , ‘ “saw. a?

‘5
s

H?"

1
m
a
.- LM A “3’3”"
>4 Is,

.. 71’: ‘
“111 ‘K 3‘
§ 3"" =3: 13%;?

h \
”C 313%! m1 I.“ 2? ‘ A; .
E 95‘7““; 7" ‘ WW1- 1 .»A. 1:1. ‘ u. -
Wyywk‘f‘ii ."C'; § .Cg . '3.“ :k ' ' '1. 1 v . u I . {I . V g
11‘ ‘ b

.i
.
Jig;
‘ - 11:13:43
,. wk“:
'1 K ..

‘1‘
WV.
12,11 1. \
Qg.i‘€:s:‘§.¥cae. L“
L; f",\’:3‘§- WM
“v 1:; .3 -

m

" 1.
‘vuw

..

n a:
.(
...

sei"

- ~r~v ‘
‘ 15'??- ,IZE'QS'
"2km M “‘"J’ 9"

 

 

*uCSiS

TYUBRARIE

Ll JolllliL

 

 

 

 

 

 

 

 

 

llllllllllllllllllllllllllllllll

3 1293 01026

 

LIBRARY
Michigan State
Unlverslty

 

 

 

This is to certify that the
dissertation entitled

ECOLOGICAL MODELING OF QUALITY OF LIFE
AMONG MIDDLE-AGED IIUSBANDG AND WIVES:
A STRUCTURAL EQUATION MODEL

presented by

Meesok Park Lee

has been accepted towards fulfillment
of the requirements for

Ph .1) degree in Family Ecology

 

 

fi/V’Mujfl/{jfifi/

jO!‘ professor

M ..
DateJfimuakf jg; / 7 $2,

MSU is an Affirmative Action/Equal Opportunity Institution 0-12771

 

 

 

 

 

PLACE IN RETURN BOX to romovo thlo chockout from your rocotd.
TO AVOID FINES rotum on or bdoro doto duo.

 

DATE DUE DATE’DUE DATE DUE

 

 

 

 

 

 

 

1 11159771322002

,__._._._1_._._ 1|": 1‘ 589-70
I @181- N23§m®
’ . p a v - u
W02 ___
“151‘! 2. 9 n," '
HUN 1 1 1997 W'CZZBKQUQZ
.0;
f y.

 

 

 

W3L

 

“ifl£&%%"

 

 

 

 

 

 

MSU leAn.“

. A ' u . . .
1 rr , ll—‘Ionl‘l

”3-9.!

 

ECOLOGICAL MODELING OF QUALITY OF LIFE
AMONG MIDDLE-AGED HUSBANDS AND WIVES:
A STRUCTURAL EQUATION MODEL

BY

Meesok Park Lee

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Family and Child Ecology

1992

ec<
he;
10:
re
in
th

ti

an
re
Ad
ti
Se

AC.

ABSTRACT

ECOLOGICAL MODELING OF QUALITY OF LIFE
AMONG MIDDLE-AGED HUSBANDS AND WIVES:
A STRUCTURAL EQUATION MODEL

BY

Meesok Park Lee

This study was designed to investigate and predict the
quality of life among middle-aged men and women in terms of
economic well-being, stress from life events, perceived
health, and satisfaction with family life, employing
longitudinal data. From an ecological perspective, the
researcher hypothesized that the quality of life of
individuals or families, assessed at one point in time, is
the outcome of the functioning of the family ecosystem over
time.

The findings of this study were based on a secondary
analysis of longitudinal data collected for the NC-164
regional research project, entitled ”Stress, Coping, and
Adaptation in the Middle Years of the Family.” Support for
the original study was provided by the Cooperative Research
Service of the U.S. Department of Agriculture, and the
Agricultural Experiment Stations of nine states. The data
were obtained from two different periods of time, 1983 and
1985, using the mail survey method.

The unit of analysis of this study was 754 men and

vo
LI
me

am

st
1i

sul

inc
out

anc‘

f an
man
eco
can
ECO
dim.
res.

Sugs

women who responded in both phases of data collection.
LISREL VII was used to test of the adequacy of the
measurement models and the hypothesized causal relationships
among selected constructs.

The results indicate that satisfaction with family life
strongly affects an individual's evaluation of quality of
life. Economic well-being, measured in objective as well as
subjective terms, is another important variable affecting
quality of life. Perceived health was a dimension which had
a significant direct influence on quality of life and an
indirect impact via economic well-being. Stress, being an
outcome, or product of interactions between the individual
and the environment, indirectly influenced quality of life.

This study demonstrated the usefulness of using the
family ecological model along with a family resource
management model to study quality of life. From an
ecological perspective, the evaluation of quality of life
can be assessed as an outcome of functioning of the family
ecosystem over time, based on satisfaction with various
dimensions of the environment. Implications for future
research and ecological theory as well as for practice are

suggested.

n-

ACKNOWLEDGMENTS

Completion of this dissertation was made possible
through the cooperation and contributions of several
individuals whom I wish to acknowledge and thank. My
dissertation committee deserves special recognition: Dr.
Rosemary Walker, Dr. Margaret Bubolz, Dr. Lillian Phenice,
and Dr. Lincoln James.

I am especially grateful to Dr. Walker, my committee
chairperson, dissertation director, and my mentor, for her
support, encouragement, and direction throughout my entire
program and dissertation process. Her standard of
excellence has been a continous source of inspiration and
challenge since I began my master's degree in 1986. I have
and will continue to learn from her high standards.

Sepecial gratitude is extended to Dr. Bubolz, principal
investigator on NC-164, for offering me an opportunity to
work on the research project. Her insight, support, and
encouragement were invaluable.

Sincere appreciation is extended to other members of my
guidance committee, Dr. Phenice and Dr. James. Their
contribution and guidance throughout my graduate program is
appreciated.

iv

.l"

 

 

 

Chi:
Offt
thr<

Ava}

sta'

the

the
exp:
Lee
dau

My:

of

sib

Thanks is extended to the Department of Family and
Child Ecology and College of Human Ecology for not only
offering a quality education, but also for supporting me
through assistantships and the Carol w. Shaffer Memorial
Award. These were all deeply appreciated.

I would also like to thank Dr. Young Kyun Lim, for his
statistical expertise, and Mr. and Mrs. Hyun K. Kim for
their support and assistance.

Finally, deep appreciation is extended to my family for
their love and support throughout my entire program. I
express heartfelt appreciation to my husband, Dr. Doo-Hee
Lee, for his strength, understanding and patience and my
daughter, Joe-Young, in enduring periods of the separation.
My parents, MI. and Mrs. Sung Yong Park, and parents-in-law,
Mr. and Mrs. Sang Yong Lee have been a never ending source
of encouragement and support. I also acknowledge my

siblings for their care of my daughter.

TABLE OF CONTENTS

page

. viii

LIST OF TABLES

ix

LIST OF FIGURES

11456689

W 1 - e'

Human Ecological Approach .

u s
e u
Income
Age .
Gender
Education .
o o ’
e .

REVIEW OF LITERATURE

INTRODUCTION

CHAPTER II

CHAPTER I

METHODOLOGY .

CHAPTER III .

Kansas Family Life Satisfaction Scale .

8 um

The Quality of Life Measure .

C

vi

In

Health Difficulty Scale

WW 1
e c ' t' S d a

esea c es' .

Research Variables . . . . .

Dependent Variable . . . .
Independent and Intervening Va
a

O O O O O O
p.

ooo§<ooooooowoooooo

c u o e t
Conceptual Model .
Hypotheses . . . .

Structural Model . . .
Overview of Analyses
Measure Validation . .
Reliability and Unidimension l
Discriminant Validity . . .
Testing the Structural Model
Model Fits . . . . . . . .

t

ooop-ooooooofloooooo
O‘

oooooooooooHoooooo
(D

ooooooooooomoooooo

CHAPTER Iv O O O O O O O O O O O O O O O O
RESULTS O O O O O O O O O O O O O O O O O O
Reliability and Unidimensionality
Discriminant Validity . . . . . .
W1 1
"Odel Fits O O O O O O
Hypotheses Test . . . .

Chapter V O O O O O O O O O O O
SUMMARY, DISCUSSION, AND IMPLICATIONS .

WW1.
Impliggtigns . . . . . . . . .

Implications for Future Research .
Implications for Family Ecological Theory
Implications for Practice . . . . . . .

APPENDICES
APPENDIX A: Table 7 Correlation Coefficients,
Means, and Standard Deviations.
APPENDIX B: Questionnaire 1983 . . . . . . . .
APPENDIX C: Questionnaire 1985 . . . . . . . .

BIBLIWRAPHY O O O O O O O O O O O O O O O O O O O

vii

38
39
39
40
40
40
41
47
47
49
51
55
55
58
58
61
63
64

68
68
68
68
75
78
78
84
89

92
92
94
100
100
101
103

106
108
118

128

Table

Table

Table

Table
Table

Table

Table

LIST OF TABLES

page

Measurement Models: Key Parameter Estimates,
t-values, Model fits, and Scale Reliability . . 69

Measurement Models After Modification:

Key Parameter Estimates, t-values,

Model Fits, and Scale Reliability . . . . . . . 73
Correlations among Constructs (0 matrix). . . . 76

Test of Discriminant Validity . . . . . . . . . 77

Structural Model: Parameter Estimates,
t-values, and Model Fits . . . . . . . . . . . 81

)8 Difference Test for Parsimonious Model . . . 90

Correlation Coefficients, Means, and Standard
DeViations O O O O O O O O O O O O O O O O O 106

viii

LI ST OF FIGURES

page
Figure 1 Ecological Model of Quality of Life . . . . . . 33
Figure 2 Conceptual Model . . . . . . . . . . . . . . . 48

Figure 3 Structural Model . . . . . . . . . . . . . . . 52

Figure 4 Measurement Model . . . . . . . . . . . . . . . 54
Figure 5 Measurement Model (I) . . . . . . . . . . . . . 71
Figure 6 Measurement Model (II) . . . . . . . . . . . . 74
Figure 7 Structural Model (with Modification). . . . . . 79
Figure 8 Parsimonious Model . . . . . . . . . . . . . . 91

ix

 

be
ec
Su
to

at‘

0V1

DUI

bei
con
and
or

liw
Wit]
by I
eval
fami
Crit

Anet}

CHAPTER I

INTRODUCTION

During recent decades, concern over quality of life has
been a growing issue, especially due to declining resources,
economic constraints, and environmental problems.
Substantial progress has been made in finding effective ways
to measure people's sense of well-being. Various research
attempts have been made to measure different aspects of life
quality and to explore how people evaluate their sense of
overall life quality, both as a whole and with respect to
numerous specific life domains.

(7 Quality of life, in general, refers to a sense of well-
bei g as perceived by people, often measured on a scale to
convey different levels of well-being. According to Bubolz
and her colleagues quality of life refers to the well-being
or ill-being of people and the environment in which they
live (Bubolz, Eicher, Evers, & Sontag, 1980). Andrews and
Withey (1976) concluded that quality of life is determined
by people's perceptions of their well-being based on their
evaluations of domains of their lives, such as marriage,
family, housing, job, friends, neighbors, and health, and
criteria, such as standards, aspirations, values, and goals.

Another study revealed that a general sense of well-being is

2
more dependent on the individual’s satisfaction with
resources than the quality of these resources (Campbell,
Converse, & Rodgers, 1976).

Some authors have defined quality of life in an
objective, status-oriented manner, examining personal
characteristics such as assets, health status, and financial
security (e.g., Gastil, 1970; Morgan 8 Smith, 1969). Others
have defined quality of life in terms of subjective
evaluation which individuals formulate about their own life
experiences (e.g., Campbell et al., 1976; Andrews 8 Withey,
1976). In this study both objective measures of resources
and status and subjective evaluations of personal life
experience are viewed as important dimensions of life
quality. Certainly, the quality of one's life appears to
encompass more than adequate material well-being. It also
includes perceptions of well-being, a basic level of
satisfaction.

In recent times many observers have reported serious
deterioration of the quality of life due to the effect of
inflation on economic well-being. Changes in the economy
and economic downturns create painful family adjustments.
The associated stress can even lead to family disintegration
(Mammen, Helmick, & Metzen, 1983). Stress has been
variously defined as a stimulus, as a crisis, as a response
to a stimulus, and as a state of the organism as it responds

to the stimulus (Dohrenwend & Dohrenwend, 1974; Hill, 1958;

3
Selye, 1956). Other researchers define stress related to
any event in which environmental demands, internal demands,
or both tax or exceed the adaptive resources of an
individual, social system, or tissue system (Monat &
Lazarus, 1977). Accumulation of stress of life events may
deplete the resources (social, emotional, or financial) that
a family can muster to meet the demands of its stressful
problems. It has been found that stress leads to
deterioration in health through depression (Brown & Andrews,
1986; Perlin, Liberman, Menaghan, & Mullan, 1981; Selye,
1956, 1980). Domains of family life are salient to the
degree of stress. Domains that have been found to influence
the level of quality of life are family life, self, family
income, and housing (Sontag, Bubolz, & Slocum, 1979).

Quality of life is an organizing concept of potential
importance to all who work toward improving the conditions
of human life: basic scientists, applied researchers and
service practitioners. In addition the broad appeal of this
concept should help to facilitate communication between
knowledge-oriented and action-oriented professionals.

Most of the research on quality of life has been
descriptive, examining factors or correlates of quality of
life for the general adult population or the aged
population. Social gerontologists have conducted
considerable research aimed at answering the question, "What

is successful or optimal aging?" (e.g., Cumming 8 Henry,

4
1961; Havighurst, 1953). Efforts have been made to describe
social patterns of aging historically. Recently, the
conditions under which aging is satisfying or debilitating
and identification of the resources and coping skills which
influence the aging process have been delineated. Most of
the research has been conducted on cross-sectional data.
These cross-sectional studies do not offer a clear picture
of the nature of family economic and psychological resources
possessed by families, nor explain how an individual's
quality of life is determined.

Knowledge of families in their middle years and the
determinants of their life quality are sparse but important
as the population ages and resources diminish. This
knowledge has lagged far behind the demand for educational
programs to address family well-being. Carefully designed
studies of the middle-years are imperative to enhance the
efforts of professionals working to increase the life
quality of families in the middle years of the family life
cycle.

W
The purpose of this research is to develop a model that

explains the relationships of selected measures to quality
of life and to predict how the quality of life is determined
in terms of family resources. Research is needed to

identify a viable model that determines the important

5

influences on the satisfaction levels of family members.

This study was designed to investigate and predict the
quality of life among middle-aged men and women in terms of
economic well-being, stress from life events, perceived
health, and satisfaction with family life, employing
longitudinal data. More specifically this study examined
how one's life quality is explained in terms of a linear
structural relationship. The term "structural" means that
the researcher will not just describe the variables or
measure association between variables but rather will
investigate an invariant "causal" relation (Bollen, 1989).

Attention was focused on people in their middle years.

B21AI1QnEhiP_QIIIDIE_§IBQX_LQ_N£:1§1

This study examines data collected for the regional
project, NC-164, supported by the Cooperative Research
Service of the United States Department of Agriculture, the
Michigan State University Agriculture Experiment Station,
and the Experiment Stations of eight other states involved.
The original study was titled "Stress, Coping, and
Adaptation in the Middle Years of the Family."

The primary focus of the regional project was stress
and coping among rural families during middle years. The
original study included both rural and urban samples, varied
employment statuses and types for men and women, and

multiple life events. Data were gathered in the area of

6
life stress, health, family financial status, marital
happiness, family happiness, and overall life satisfaction.
The scope and depth of the data facilitate the examination
of perception of individual life quality among middle-aged

people.

WW
Human Ecological Approach

This study is based on the human ecological framework.

_/

15\A human ecological approach is a holistic approach to
examine the interdependence between humans and their
environments.! For example, it examines the
interrelatiogships between family members, between the
family and their work and other environments, and the socio-
economic system. The human ecological framework enables the
researcher to examine phenomena in the wholeness of
interaction and interdependence (Bubolz, Eicher, & Sontag,
1979). Because humans perceive their environment, make
decisions, and satisfy needs through interaction, the
phenomena must be examined from the stand point of wholeness
and in terms of interaction and interdependence.

\XA human ecological framework provides a basis for
delineating various kinds of quality of life indicators,
defined as indices or measurements of various aspects of

human life or environmental conditions.\ These tell

—4

something about the degree of ill-being or well-being of

7
human conditions or environments.

In the human ecological framework, the individual and
the family unit are nested in three interrelated
environments: the natural, biological, and physical
environment; the human built environment; and the social-
cultural environment (Bubolz & Sontag, in press). The basic
components of the human ecosystem serve as a guide in
clarifying what phenomena need to be described and what data
need to be selected as indicators. Indicators can measure
something about the conditions or status of people (the
human environed unit); they can describe conditions or
resources of the environments (natural, human constructed or
human behavioral); or they may explain something about
people's interaction with the environment and their use of
resources. A human ecological framework provides a means
for conceptualization, measurement, and analysis of the
quality of life (Bubolz et al., 1980).

In the human ecosystem model, it is assumed that humans
are a part of the total life system. Therefore, humans
cannot be considered apart from all other living organisms
in the environment. The environments for the family provide
the resources necessary for life and constitute a life
support system (Andrews, Bubolz, & Paolucci, 1980). Family
members transform energy, matter, and information from their
environment to meet their needs.

In the human ecological framework, quality of life is

8
defined as perceived satisfaction with life as a whole.
This satisfaction results from: (1) the interaction of
individuals possessing certain needs and characteristics
with resources available in their various environments, and
(2) the degree of person environment fit. Therefore, in
this context, it is hypothesized that quality of life is
influenced by the degree of satisfaction with selected life
concerns. These concerns are indicated by variables which
represent human needs, conditions, and resources related to
the natural-biological, human constructed, and behavioral

environments.

QQDE§P§931_HQ§21

Components of the Deacon 8 Firebaugh Family Resource
Management Model (Deacon 8 Firebaugh, 1988) are being
utilized here. Deacon and Firebaugh conceptualized family
resource management as a system comprised of inputs
consisting of demands and resources, throughputs of planning
and implementing, and outputs of met demands and used
resources.

This study focuses on the effect that inputs (demands,
including life events, goals and resources) have on quality
of life as an output. Demands are represented by stress
caused by stressors that relate to the economic environment,
time, the family environment, and health. Resources are

represented by health and economic well-being as measured in

9
objective (e.g., per capita income) and subjective terms
(e.g., perceived income adequacy). Satisfaction with family
life, and overall quality of life are assumed to be family
goals. Respondents' evaluations of their family life and
quality of life represent the degree to which these needs

are met and evaluated as part of the system's output.

I' '! !° 1 E Ii
Some limitations are inherent in this study. The

limitations relate to the characteristics of the subjects
and the method in which the sample was collected.
Generalizability is limited due to the imbalance of race and
income. The sample is predominantly Caucasian, and over-
represents higher income and higher education groups, as
well as the rural population. The original study intended
to examine stress in intact families. Therefore the sample
was purposely limited to married persons with at least one
child. The sample was randomly drawn using mailed-
questionnaires. The questionnaire dealt with sensitive
areas of finances and family interactions. The sample may
under-represent people who are less satisfied with their
overall lives. In general, it is recognized that the survey
method can not directly observe behavior nor can it measure
the phenomenon in great depth.

The following assumptions are identified with regard to

the research study.

10

1. The participants of this project responded to the
questions on the research instrument to the best of their
awareness as completely and fully as possible to the limits
of their consciousness. It is also assumed that they
completed the questionnaire independently, not being
influenced by other persons.

2. Quality of life is normative. A high quality of
life is a goal of individuals. It must be based on what is
desirable and what is possible for the well-being of the
individual.

3. Quality of life is affected by multiple indicators,
recognizing that individual and family life encompasses
several interrelated dimensions.

4. Quality of life is best explained by incorporating
objective and subjective indicators of environmental
conditions and resources as well as of individual and family
well-being.

5. Individuals experience stress as a result of life
events from total environments. Although the degree of
stress may be different, all individuals experience stress.

6. Individuals experience stressful life events in
their everyday life, including family life.

7. Stress is a potentially negative influence on an
individual's health as well as psychological, emotional,
physical, and social satisfaction.

11
8. The middle years are a period with some unique life
events. It is also assumed that these life events do not
necessarily result in a higher degree of stress because they

have been experienced in the middle years.

CHAPTER II

REVIEW OF LITERATURE

Related research and literature are reviewed in six
areas. These are quality of life, quality of family life,
health, stress and life events, economic well-being, and

middle years of life.

W

Two major studies on quality of life were conducted in
the early 19708 (Andrews 8 Withey, 1976; Campbell et al.,
1976). These were done by the Survey Research Center of the
University of Michigan. Employing a national sample, both
studies examined how people's sense of well-being was
related to different aspects of life.

Both studies indicated that an individual's overall
satisfaction with life is based on the satisfactions related
to various life domains. Andrews and Withey (1976)
classified life concerns into domains and criteria, and
further concluded quality of life is determined by people's
perceptions of their well-being based on their evaluations
of domains and criteria. Domains include places, things,
activities, people, and roles that are part of people’s

lives and their environment. They include marriage, family,

12

 

s:
01
an
he
3P
sp;

na1

sat
ref
suc

act

that
for

and

Pers
dOma‘
nest
and t

13
housing, jgbmeriend , neighbors, and health. Criteria are
standards, aspirations, values and goals. People judge or
make evaluations based on the criteria and domains of life.
Andrews and Withey found the most important domains which
contributed to quality of iife were satisfaction with self,
satisfaction with family fife, satisfaction with the amount
of fun in people's life, and satisfaction with family income
and consumption. The ddmains contributing less were
health, job, goods and services, residence of house or
apartment, things to do with family, time to do things,
spare time activities, and an index of satisfaction with

national government.

’ ' a r’ *- '
i . ’1)?
d One—N

Andrews and Withey have further demonstrated that
satisfaction with one's life as a whole is additive. It
reflects the sum of one's satisfaction in various domains,

I

such as one’ s incomeA,lh6using, occupation, leisure
activities,‘;;;IIyM life.

The researchers of the two major studies have found
that people reserved their ratings of greatest satisfaction
for those parts of their lives that are the most personal
and intimate. The more distanced a domain was from the
person, the more dissatisfied they are likely to be with the
domain. Campbell and his associates found Americans are
most likely to express high satisfaction with their marriage
and their family life and low satisfaction with their

economic status and their education. Satisfaction in the

14
domains such as self, standard of living, family life,
marriage, friends, and work has the greatestwinfluence in

accounting for the level of overall quality of life.

IDOO‘O

":

A great deal of research has shown that a positive /
relationship exists between income and qualityi;£:;ife. /
Many studies have found that objective income {9 correlated
with quality of life (Alston, Lowe, 8 Wrigley, 1974; Andrew
8 Withey, 1976; Mancini 8 Orthner, 1980; Riddick, 1980).

The studies concluded that a relationship exists even when
other variables, such as education, were controlled.
However, the effect of income was reduced when other factors
which interact with income, such as better health, were
controlled.

As might be expected, subjective income measures, such i
as satisfaction with income, and perception of income 7'

adequacy, were also related to quality of life (Ackerman 8

Paolucci, 1983; Berry 8 Williams, 1987; Braun, 1977;

 

Campbell et al., 1976; Spreitzer 8 Snyder, 1974). Spreitzer
and Snyder (1974) found that people's perception of
financial adequacy was a substantially stronger indicator of
life satisfaction than objective economic indicators.

Other researchers have found that some subjective
indicators make a larger contribution and are better

predictors of people’s evaluation of the quality of life

15
than objective indicators (Ackerman 8 Paolucci, 1983; Berry
8 Williams, 1987; Campbell et al., 1976). On the basis of
several studies Campbell (1981) concluded that objective
conditions and factors contribute to well-being to the
degree that they provide positive and satisfying experiences
and contribute to people's subjective evaluation of their
lives and certain domains of their lives, such as the
family.

In their study of quality of life, Ackerman and
Paolucci (1983) examined subjective and objective income
adequacy. Objective income measurements were calculated and
adjusted for family size and place of residence. Subjective
measures of income adequacy were measured by the degree to
which the family felt satisfied with its income being enough
to live on. The results indicated that people's evaluations
of the quality of life were positively correlated with their
subjective as well as objective income. The study also
found that subjective indicators of income adequacy
explained more variation in quality of life. The study
concluded that individual evaluations of income relate to
the societal average and satisfaction depends on how an

individual perceives his/her income to match the average.

Ago
Studies that include age variables have produced

contradictory findings. In the 19603, some studies found

16
that young people were happier than their elders (Bradburn 8
Caplovitzs, 1965; Gurin, Veroff, 8 Field, 1960). In other
studies, however, no age effect was found (Andrews 8 Withey,
1976; Spreitzer 8 Snyder, 1974). Several other studies
found a positive relationship between age and quality of
life (Braun, 1977; Campbell et al., 1976; Cantril, 1965;
Medley, 1980). Braun (1977) found that younger people
reported higher levels of both positive and negative
affects, but older groups reported greater levels of overall
quality of life. Given the confusing nature of
contradictory findings, a meta-analysis using studies done
prior to 1980 revealed that the correlation between age and
quality of life was near zero, regardless of whether other
variables were controlled(Stock, Okun, Haring, 8 Witter,
1983).

In a large scale study, Campbell and his colleagues
(1976) reported that satisfaction with life and their Index
of General Well-Being correlated positively with age;
whereas reports of being very happy decreased with age. The
study also found that older persons reported greater
satisfaction in every domain except health. Host results
show a slow rise in satisfaction with chronological age;
however, it seems that younger people tend to experience
higher levels of satisfaction while older people tend to
judge their lives in more positive ways.

In recent years, life stages that create

17
characteristics and demands have been examined (Estes 8
Willensky, 1978; Harry, 1976; Medley, 1980). Employing
multivariate analysis on a national probability sample,
Medley (1980) found that life satisfaction was high for men
and women at each stage across the life span; four stages of
adulthood were early adulthood (22-34), early middle age
(35-44), late middle age (45-64), and late adulthood (65 and
over). When gender was combined with the life cycle
variable an interesting finding resulted. Men showed a
monotonic increase in life satisfaction as their life cycle
stage increased; while women showed a relatively constant

relationship over the life span.

Gender

Little difference has been found between the sexes in
the evaluation of overall quality of life (Andrews 8 Withey,
1976; Campbell et al., 1976; Goodstein, Zatra, 8 Goodhart,
1982; Olsen, 1980; Toseland 8 Rasch, 1979-1980). However
some studies found a gender age interaction; younger women
were happier than younger men, while older women were less
happy than older men (Medley, 1980; Spreitzer 8 Snyder,

1974).

Education
Studies show education has an influence on quality of

life, although the effects of education seem weak and may

18
interact with other variables such as income (Bradburn 8
Caplovitz, 1965; Campbell, 1981; Palmore, 1979). However,
several other studies showed no significant effect of
education on quality of life when other factors were
controlled (Spreitzer 8 Snyder, 1974; Toseland 8 Rasch,
1979-1980). Some studies found that education has a more
positive effect on quality of life for women (Freudiger,
1980; Glenn 8 Weaver, 1981; Mitchell, 1976). According to
Campbell (1981), education may serve as a resource for a
person and may also raise aspirations and alert people to

alternative types of life.

Quali§¥_2f_£amil¥_hife

Several studies found that satisfaction with marriage
and family life is the strongest predictor of quality of
life (Andrews 8 Withey, 1976; Bubolz, Eicger, Evers, 8

I

Sontag, 1980; Bubolz, Walker, 8 Lee, 1989;\Campbell, 1981; E
Campbell et al., 1976; Freudiger, 1980; Gfienn, 1975; Glenné
8 Weaver, 1979, 1981; Haavio-Mannila, 1971; Michalos, 1980,
1982; Sontag, Bubolz, & Slocum, 1979; Toseland a Rasch,
1979-1980; Walker, Lee, Bubolz, 8 Keefe, 1990; Walker,
Bubolz, 8 Lee, 1991). Glenn (1975) reported that although
married women reported greater stress symptomsfithan their
unmarried counterparts, they also tended to report a higher
quality of life. Glenn 8 Weaver (1979) found that ‘\\ W

satisfaction with marriage was the strongest determinant of

"‘M
*H mm

19
quality of life when other factors such as income,
education, and occupation were controlled.

According to Glenn and Weaver (1981), previous divorce
is not related to the happiness of persons who are
remarried. The researcher concluded that marriage has an
effect on people's evaluation of quality of life; it was not
simply a selection factor of happier people getting married,
or staying married. In contrast others found that when
people evaluate their quality of life, satisfaction with
marriage and family life is more important than the
objective fact of marriage (Andrews 8 Withey, 1976;
Campbell, 1981; Campbell et al., 1976; Glenn 8 Weaver, 1979,
1981).

Recent research on quality of life identified major
domains which have had great impact on quality of life
(Bubolz et al., 1989; Walker et al., 1990, 1991). It was
found that life quality is determined not only by objective
conditions and resources but also by the subjective
evaluations of these conditions in relation to demands and
aspirations. Among those the assessment of life related to
family relationships was the strongest predictor for quality
of life. Several studies found negative effects of having
children on quality of life (Andrews 8 Withey, 1976; Glenn 8

McLanahan, 1981; Glenn 8 Weaver,1979).

20
39.11321

Based on a national sample, Campbell (1981) concluded
that poor health was associated with dissatisfaction with
health. The researcher expected poor health to be
associated with a degree of dissatisfaction in other domains
of life. It was found that people with serious disabilities
were in fact considerably more likely than people with no
health problems to express dissatisfaction with their lives.

Results of several other studies found a significant
relationship between self-rated health and quality of life
(Edwards 8 Klemmack, 1973; Larson, 1978; Markides 8 Marti ,
1979; Near, Rice, 8 Hunt, 1978; Riddick, 1980; Spreitzer 8
Snyder, 1974; Toseland 8 Rasch, 1979-1980). The self-rated
health variable was measured by respondents rating their
health as excellent, good, fair, or poor. This effect
remained when other variables such as socio-economic status
(SES) and age were controlled (Clemente 8 Sauer, 1976;
Freudiger, 1980; Larson, 1978).

In a large-scale study, Campbell and his associates
(1976) found that although health was rated by subjects as
the most important factor in happiness, satisfaction with
health was only the eighth strongest predictor of life
satisfaction. On the other hand, although some other
studies found a strong zero-order correlation between health
and quality of life, they found no relationship when other

factors such as leisure activities were controlled ( Mancini

21
8 Orthner, 1980; Miller, 1980). This indicates that part of
the influence of health on quality of life is not simply the
direct effect of how people feel about their health, but
also on what their health allows them to do.

Many studies on health and life quality used more
objective measures of health such as disease checklists
(Larson, 1978; Liang, Kahana, 8 Doherty, 1980; Mancini 8
Orthner, 1980), while some other studies found physicians’
ratings also correlated with quality of life (Palmore 8
Luikart, 1972; Larson, 1978). Perceiving oneself as being
in good health was found to be positively related to
satisfaction with life, while the number of experienced
ailments, either recent or immediately past ones, were
unrelated (Edwards 8 Klemmack, 1973).

According to Zautra and Hempel (1983), it appears that
subjective health measures have a stronger relationship to
quality of life than objective measures of health, although
both measures were significant. However, one study showed
that health influenced people's quality of life only cross-
sectionally, not longitudinally (Miller, 1980). This
finding raises questions about the process and causal
direction by which health and quality of life are related
(Zautra 8 Hempel, 1983).

Other studies found that poor sleep has been related to
lower quality of life (Roth, Kramer, 8 Roehrs, 1976;

Sherman, 1980; VanCoevering, 1974). The direction of

22
causality is not clear because distressed persons are more

likely not to sleep as well as others.

W

An extensive body of research has found a negative

-- —.-...—-_

impact of stress on overall quality of life (Caplan, 1983;

‘~\

Holmes 8 Rahe, 1967; Perlin et al., 1981; Selye, 1956,
1980). Some researchers have obtained consistent results
related to the impact of life events on quality of life
(Kammann, 1982; Miller, 1980). Good events were related to
positive affect, and bad events were related to negative
affect (Reich 8 Zautra, 1981, 1983; Warr, Barter, 8
Brownbridge, 1983; Zautra 8 Reich, 1980). Reich and Zautra
(1981) hypothesized that well-being was composed of two
factors, a positive component and a negative component.
Demand measures were positively related to the negative
component, lower quality of life; desires were predicted to
relate positively to the positive component. In their study
the concepts of demands and desires were developed to
capture the essential distinction between daily activities
which occur because people want to do them and those which
arise from the pressures of living. The impact of life
events such as the death of a loved one or relocation of
family members on quality of life has been found (Dohrenwend
8 Dohrenwend, 1974; Vinokur 8 Selzer, 1975; Williams, Ware,

8 Donald, 1981).

23

In a study on family life events which were more
frequently reported and perceived as stressful, Williams and
Bubolz (1986) reported that some family life events were
perceived as very stressful. They were death of family
member, family pet dies, member demands new privileges,
household chores pile up, death of close friend or
confidant, member experiences serious emotional problems,
and family member involved in the courts.

Studies have found that people under stress are more
susceptible to illness, depression, anxiety, low self-
confidence, and dissatisfaction than are people who are not
experiencing stress (Caplan, Cobb, French, Harrison, 8
Pinneau, 1980; Kahn, Wolfe, Quinn, Snoek, 8 Rosenthal, 1964;
Rabkin 8 Streunig, 1976).

In a study on psychological concepts related to
perceptions of life quality, Abbey and Andrew (1986)
concluded that stress and depression relate strongly and
negatively to perceptions of life quality. The results
indicate that individuals who feel under stress or depressed
report feeling worse about their lives than do unstressed or
happy individuals.

Other research evidence shows that one's ability to act
and control events is related to quality of life (Guttmann,
1978; Reich 8 Zautra, 1981). Although earlier research
studies have shown that whether an event is controllable is

an important factor on people's evaluation of quality of

24
life, an understanding of the impact of large scale events
and of the cumulative impact of daily events seems to be

needed.

WM

Economic well-being is one of the indicators used in
the study of quality of life. Economic well-being has been
measured in both objective and subjective terms. Objective
measures include variables such as per capita household
income, while subjective indicators include measures such as
income adequacy and satisfaction with resources. Objective
measures do not take into account quality measures and
mental processes important to individual perceptions. In
addition these do not take into account all of the
individual and family differences which may be pertinent.
Likewise, some limitations of subjective measures have been
recognized. Fletcher and Lorenz (1985) indicated that use
of only subjective measures may be misleading if individuals
have become resigned to their level of income and adapted to
it.

Recently, authors have recognized the importance of
both measures in the study of economic well-being and
provided evidence that the combined measures are better
predictors than one measure used alone. Ackerman and
Paolucci (1983) studied subjective as well as objective

income adequacy as related to overall quality of life.

25
Objective income was measured by comparing the reported
income to the moderate income of the Bureau of Labor
Statistics budgets adjusted for the family size and area of
the country. Perceived income adequacy was measured by
asking the degree to which respondent felt their family
income was enough to live as comfortably as they liked.
Using a multistage probability sample of private households,
the researchers found that life quality measures of
satisfaction increased with an increase in either subjective
or objective income adequacy. They also found that the
subjective measure accounted for more variation in
satisfaction with income and overall life quality than
objective adequacy.

Hafstrom and Dunsing (1973) examined both subjective
family income adequacy and income before taxes, unadjusted
for family size, to satisfaction with overall quality of
life. Using a sample of homemakers, they found subjective
adequacy to have the highest correlation with life
satisfaction; income before taxes had a smaller but moderate
correlation with life satisfaction. They also found that
subjective adequacy predicted more of the variation in life
satisfaction than did the objective measure. These
researchers further suggested that the two kinds of measures
should be used as compliments to each other, for the
combined measures are better predictors than either type

used alone.

26

In a study on economic well-being, Fletcher and Lorentz
(1985) found a less strong relationship between subjective
income and objective income for older respondents than for
younger ones. They used total family income and
satisfaction with family income related to economic well-
being for different age, sex, and race groups. Satisfaction
with income increased with age without an actual
corresponding increase in total family income. A positive
relationship between family income and satisfaction with
family was found in both female and male groups. Overall,
nonwhites were less satisfied.

Satisfaction with one's financial situation is an
important factor in an individual's sense of economic well-
being. In a study of factors having an impact on financial
satisfaction, Davis and Helmick (1985) used measures of an
additive index including satisfaction with consumption
level, satisfaction with wealth, and satisfaction with
financial security. The measures of financial security
included these satisfactions: the proportion of income
saved, emergency reserves, and future financial security.
Other variables included family income, family net worth,
debt to income ratio, remaining number of child-rearing
years, and number of earners in the family. The researchers
also included desire for financial improvement and perceived
change in financial condition. Employing multivariate

analysis for six groups (husbands and wives from three

27
states), the researchers concluded that desire for financial
improvement and perceived change in financial improvement
was a strong predictor of financial satisfaction in all six
groups. Net worth was also found to be a strong predictor
in four of the groups. However, family income was not a
strong predictor in any of the groups. The findings are
consistent with previous findings (Hafstrom 8 Dunsing, 1973;
Yaar, 1976) that objective indicators such as family income
or net worth were significant measures of financial
satisfaction. The best predictors, however, seemed to be
perceived change in financial condition and desire for
financial improvement. It was further concluded that both
were significant predictors of financial well-being.

One study found that families tend to judge their
current economic status on the basis of past economic
conditions as well as their future expectations (Moen, Kain,
8 Elder, 1983). Changes in environment external to the
family (inflation, recession, or war) affected how the
family felt about its security. Fear of job loss or concern
that current income would be inadequate were related to
perceptions of decreased economic well-being although income
was unchanged. Changes internal to the family (illness of
family member, change in family composition) may have the
same effect. These may result in increased demands or
decreased resources which may further result in a sense of

decreased economic well-being (Davis 8 Helmick, 1985;

28

Hafstrom 8 Dunsing, 1973; Yaar, 1976).

W

In recent years attempts have been made to develop
sociopsychological conceptions of the life course and its
various developmental periods (Hunter 8 Sundel, 1989).
Studies based on the theories of life cycle have become
important. The increase in longevity has brought more
emphasis on the midlife period because this period has
lengthened. According to Schaie and Willis (1986), the
average couple can now expect to live almost twenty years
alone in an "empty nest" after the last child has left home.
Midlife has become a typical stage most individuals will
experience. Until the mid 1970's, little research focused
on the middle years. ,

The definition of middle years differ by researchers.
Some use chronological age boundaries for life stages.
Among them, Levinson defined the midlife era from 35-40 to
60-65. But others view middle age as a state of mind
(Fiske, 1979) or as a period during which individuals create
highly complex and demanding roles when compared to other
stages of the life cycle (Cohler 8 Boxer, 1984).

Levinson (1978) has emphasized the importance of the
midlife transition (midlife crisis). Campbell and his
colleagues (1976) found that transitions similar to those

described by Levinson do occur in the lives of most men and

29
women in American society. They found that there are long
term patterns through the life span.

Individuals' perceptions as to whether they are young,
middle aged, or old are strongly influenced by their
particular social contexts, such as gender, social class,
health, nutrition, beliefs, goals, and social values (Hunter
8 Sundel, 1989). One theory of human development is to view
it as socially organized and socially produced, not only by
what happens in the early stages, but also by the effect of
social structure, and social interaction, and their effects
on life chances throughout the life course (Brandtstadter,
1984). Social structure is considered powerful, because it
shapes one's beliefs about aging and one's appraisal of life
events.

Researchers have found a positive association between
age and expressed satisfaction with a number of areas of
life (Campbell et al., 1976). Based on their 1971 Quality
of Life study, the researchers found that people in older
stages of the life cycle report somewhat higher
satisfactions than their younger counterparts. ,They
explained that the higher satisfaction among older adults is
a function of their declining aspirations.

One approach to understanding how people evaluate their
quality of life has focused on the "gap" between what people
have and what they aspire to (Andrews, 1981; Mason 8

Faulkenberry, 1978; Michalos, 1980, 1983). The key notion

30
is that people whose actual life conditions and aspirations
are closely matched will assess their life quality as much
higher than will people for whom there is a large gap
between actuality and aspirations.

Studies have found that people in later life stages
experienced a decline in stress due to fewer life events. A
typical life event scale, often used to assess the level of
stress, showed a declining number of such events with
increasing age. Campbell and Converse (1980) included
thirty life events in the 1978 Quality of Life survey. They
concluded that, with increasing age, respondents reported a
decreasing number of life events for the previous five
years.

Some researchers have recently argued that the
traditional major life events methodology may not be the
best assessment of stress because stress may be created
through "daily hassles” that people experience when carrying
out routine transactions within their environments (Kanner,
Coyne, Schaefer, 8 Lazarus, 1981). The number of these
daily hassles probably declines for older persons, because
they are released from many of the roles and constraints
that make life difficult. The stress that arises in the

middle years may be unique to the time period.

31

WW
An Application of the Human Ecological Perspective to

Quality of Life

As indicated in Chapter I, ecological theory is the
study of the interrelationships of organisms and
environments. In this perspective, it is critical to
realize that human beings interact with the total
environment. It is further recognized, in the ecological
perspective, a change or action in one part of the ecosystem
induces a change in another part (Bubloz 8 Sontag, in
press). According to Bubolz and Sontag, mutual or
reciprocal interaction occurs when any part of an ecosystem
causes a change in any other part and, in return, is

changed.

._\-’

I!
">4"
, .

rFrom this ecological perspective, the.rg§earcher

hypothesizes that the quality of life of individuals or
families,<assessed at one point in time), is the outcome of
the functioning of the family ecosystem over time. VThe
assessment of individual-or.family,quality.ofwlife can be
made based on/objective and subjective satisfaction with the
'Jquaiity of the natural physical biological, human built, and
social-cultural environments and the resources which come

from the environments or from family members _themselves;/

;
/

In characterizing the relationships in an ecosystem,

Griffore and Phenice (1988) emphasized multidimensional and

32

reciprocal causality. Consistent with this view, the
present investigator proposes a multidimensional measure of
quality of life and will examine possible reciprocal
causation of selected variables over time. In this study,
the interactive nature of family processes and relationships
with environments (crucial characteristics of an ecosystem)
can be captured by utilizing Linear Structural Relation
analysis with longitudinal data.

Based on the ecological theory, an ecological model of
quality of life is presented in Figure 1. An individual's
assessment of quality of life today is the outcome of past

events and conditions.

33

QUALITY OF LIFE

PRESENT QUALITY OF FAMILY LIFE

a

STRESS HEALTH HELL-BEING

N

 

 

 

 

 

\/

QUALITY OF EAHILY LIFE

 

 

PAST STRESS * TE

(From total environments) (Resource)

Figure 1 Ecological Model of Quality of Life

CHAPTER III

METHODOLOGY

The primary foci of this investigation were to examine
predictive factors and to formulate an ecological model of
quality of life over time. The study was designed to
investigate and predict the quality of life among middle
aged individuals in terms of stress, economic well-being,
perceived health, and quality of family life, employing
longitudinal data.

The data utilized in this study were collected for the
NC-164 regional research project entitled "Stress, Coping,
and Adaptation in the Middle Years of the Family." Support
for the original study was provided by the Cooperative
Research Service of the U. S. Department of Agriculture, and
the Agricultural Experiment Stations of nine states:
Indiana, Iowa, Kansas, Kentucky, Louisiana, Michigan,
Minnesota, Missouri, and Nebraska. The following section
describes the research design and subjects, research

variables, hypotheses, and analysis of data.

34

35
WW
Core Study
This study is based on a secondary analysis of
longitudinal data from the NC-164 regional project. Data
were gathered over two different time periods, 1983 and
1985, using the mail survey method. The study examined
husbands and wives as they existed at each point in time.
Additionally, the 1983 study asked questions on the family's
experiences over the past three years. In 1985, those same
individuals who responded in 1983 were asked to give

information about their experiences over the past two years.

W

The sample for project NC-164 was randomly selected
from a marketing corporation's list of names and addresses
of intact families in the middle years in the nine states
participating in the study. A stratified sampling design
was employed. Each state selected a Standard Metropolitan
Statistical Area (SMSA) with a population of 20,000 or more
as the site for the urban sample. The rural sample was
drawn from a county not part of or adjacent to a SMSA.
Sampling areas were selected to be representative of the
state in terms of median family income and educational
attainment of persons age 18 and older. In the project,
middle-stage families were defined as families in which the

wife was between the ages of 35 and 65 with at least one

36
child. The rural sample was over-represented because of a
primary interest of the research team in rural families.
Response rates by states in 1983 ranged between 30 percent
and 40 percent, considered satisfactory in light of the
length and complexity of the questionnaire. The overall

response rate for 1985 was a little over 50 percent.

EIQEQQBI§_H§§§
The questionnaire for project NC-164, developed by the

research team from the participating states, primarily
focused on stress among families during mid-life. In
February 1983, each wife and each husband were mailed
separate packets, containing a letter of introduction, a
questionnaire, and a postage-paid return envelope. This
method was employed to insure independent completion of the
questionnaires. Follow-up post cards and a second set of
questionnaires were mailed to encourage broader
participation. In March 1985 all participants who had
responded in 1983 were mailed similar packets. An identical
procedure was used in March 1985. Approval was obtained
from the University Committee on Research Involving Human

Subjects for all research procedures and instruments.

W
The research instrument used in this study was

developed by committee members of the Agricultural

37
Experiment Station Regional Research Project NC-164, North
Central Region. The instrument, a self administered mail
questionnaire, included information representing the varied
interests of the research team.

The instrument used in 1983 was thirty-two pages in
length and took about one hour to complete. The instrument
was pre-tested twice in the development process. As part of
the development process, each state pre-tested the
instrument and then pooled the information to identify
problems and make appropriate changes.

A similar research instrument was used for the 1985
collection of data. In the 1985 instrument, a few questions
were added concerning changes in family life events and
economic status. Some questions were revised to make the
instrument clearer. Questions which were confusing or not
central to the overall purpose of the project were
eliminated, resulting in a shorter questionnaire of 23

pages.

The Quality of Life Measure

Overall quality of life was measured using the
Campbell, Converse, 8 Rodgers (1976) Index of Well-being.
Campbell, Converse, and Rodgers developed the Index of Well-
being using the Index of General Affect (multi-item general
use scale) in a large scale study of The Quality of American

Life (1976). The Index consists of respondents' rating of

38
their lives on an eight semantic differential scale
concerning feelings about present life (eg., a scale of 1 to
7 relative to life being boring-interesting, miserable-
enjoyable, and useless-worth while) and life as a whole.
Evaluation of life as a whole was measured on a 7 point
scale ranging from completely satisfied to completely
dissatisfied (Appendix C, 957, 058a, p127). The reliability
of this index was estimated as .89 in Campbell’s study

(Campbell et al., 1976).

Kansas Family Life Satisfaction Scale

Quality of family life was measured by a four-item
scale dealing with satisfaction in family relationships: (1)
relationship with spouse, (2) relationship with children,
(3) childrens' relationships with each other (4) family
relationships in general. The scale was developed by Schumm
and his associates (1986). (Appendix B, Q43, p117). The
scale was found to be substantially correlated with quality
of life in their study of 620 families in twelve states (r =
.46 for mothers, p < .001). In the study the scale proved
to have acceptable internal consistency and reliability,

above the .80 criterion.

Health Difficulty Scale
The health difficulty scale (Appendix B, 018, p115) was

a twelve-item scale of general difficulties which can be a

39
health difficulty by itself or be an indicator of another
illness. Respondents were asked to rate the frequency with
which each item occurred on a five point scale ranging from
1 (never) to 5 (almost always). The points were added for
the 12 items, the higher the score the more health
difficulties. This scale, and all succeeding scales, were

the same at both times, 1983 and 1985.

WWW

McMahon (1988) describes the characteristics of the
1983 core sample. The mean ages for women and men were 45.1
and 48.1 respectively (Williams 8 Bubolz, 1986). The mean
number of years married to the present spouse was 24. In
1983, 61 percent of families had two or three children,
while 34 percent had four or more and 5 percent had one
child. The sample was 95 percent white,and 72 percent of
the sample resided in rural areas in contrast to 28 percent
from urban areas. In religious preference, 67 percent women
and 63 percent men were Protestant, while 23% and 30% men
were Catholic, and the remainder "other.” The majority of
men were employed full-time; women were evenly distributed
among full-time employed, part-time employed, and full-time
homemaker. The mean years of education of women and men

were 12.9 and 13.1 respectively.

40
W

The present study utilizes data for 754 married women
and 754 married men who responded to both phases of data
collection, 1983 and 1985. The respondents were from seven
states: Iowa, Kansas, Kentucky, Louisiana, Michigan,
Missouri, and Nebraska. Data from Indiana and Minnesota
were eliminated due to inconsistencies in coding and data
handling.

In 1985 two-thirds of the sample lived in rural areas
or towns with population less than 50,000. Mean age of
women was 53 and that of men was 56. Women's mean years of
education was 13 years; men's was 14 years. Median family

income was between $35,000 and $40,000.

W
In summary, the major purpose of the present study is

to investigate the selected factors that cause various
assessments of life quality among individuals. The model is
developed based on the human ecological theory. This study
is longitudinal in design involving two phases of
assessment: 1983, 1985. The unit of analysis is 1508
married individuals from whom completed questionnaires were

available for the two assessment periods.

41
Esseamhlaniables

Dependent variable
W

ggnggptg§1_pgfiinitign: Quality of life refers to an
individual's evaluation of overall life satisfaction. For
this study, it is defined as a person's sense of well-being,
his/her satisfaction or dissatisfaction with life, or
his/her happiness or unhappiness.

ngratigngl_ngfiinitign: Quality of life was measured
using the Index of Well-being (Campbell et al., 1976). It
is an eight item semantic differential scale concerning
feelings about present life and one question about life as a
whole. On a scale of one to seven, respondents were asked
to describe their feelings on each items. (Refer to Appendix
C, 057 and 058a, p127). The score to obtain the quality of
life measure averaged the individual scores of the eight
statements and added the weighted score (1.1) of feelings
about satisfaction with life as a whole. In calculating
this score the procedure of Campbell and his associates
(1976) was adopted. The possible range was 2.1 to 14.7. In

the analysis the response from 1985 was used.

Independent and Intervening variables

W
Congeptua1_ngfinitign: The level of satisfaction one

receives from the relationships within one's nuclear family

42

composed of husband, wife, and children.

ngzgtigngl_2gfiinitign: The quality of family life was
measured by questions used in the Kansas Family Life
Satisfaction Scale (Schumm, McCollum, Bugaighis, Jurich, 8
Bollman, 1986), consisting of four statements regarding
satisfaction with family relationships. The four statements
asked respondents how satisfied they were with family life
(SFL83,SFL85), relationship with spouse (SSP83, SSP85),
relationship with children (SCH83, SCH85), and the
relationship the children had with each other (SCE83,SCE85).
They were to rate each statement on a scale of 1 to 7 scale
(completely dissatisfied to completely satisfied). The 2
variables, QFL83 and QFL85, are based on data from the 1983
and the 1985 surveys, respectively (Appendix B, Q43, p117,

and Appendix C, Q58, p127).

29I21122Q_E2£1&h_ifiLIE§§L_ELIE!£1

ggnggptgal_ngfini§19n: It is defined as the level of
objective and subjective health as perceived by the
individual.

ng:3§1gna1_figfiinitign: In 1983, respondents were
asked to score 12 statements related to health problems or
health habits. The statements concerned: had trouble
sleeping, had accidents, been irritable, smoked cigarettes,
used prescription drugs, had a weight problem, used alcohol,

found it difficult to relax, had muscle tension, had nervous

43
indigestion or anxiety, had cold or flu, had headaches, and
had been depressed. Each item for the statements had a
possible score of 1 to 5 (never to always). The scores were
recoded for the twelve items (1 - always, 5 = never) and
then added together to create HLTH83. HLTH85 was developed
in an identical way from the 1985 data. A higher cumulative
score represented fewer symptoms or problems and a higher
level of health than a lower cumulative score. The
perceived health variable could range from 12 to 60

(Appendix B, Q18, p115 and Appendix C, 027, p124).

fii!1£l.1§13§§21.§13§§§1

anggptggl_ngfin1tign: Stress can be defined in
various ways based on the context in which it is being

employed. According to Antonovsky (1980), stress is the
loss of desired homeostasis that requires energy which is
not readily available. Stress can cause disruption of the
social unit or system (Monat 8 Lazarus, 1977). Stress for
this study is defined as a personal reaction to an
experience which puts demands on the individual.
ngrgtigna1_ngfiinitign: Stress was measured by the
cumulative scores on the Family Life Event Scale, an
adaptation of the Family Inventory of Life Events (FILE),
which was developed to measure family stress (McCubbin,
Patterson, 8 Wilson, 1987). It was the measure of

collective life events including internal events related to

44
family transitions and relations, finances and health, as
well as interactions with events in external environments
such as school, work, and the law. In 1983 respondents were
asked to indicate if certain events had occurred in their
families within the last three years. In 1985 they were
asked if the events had occurred within the last two years.
The following coding method was used. If the event had not
happened to the respondents, the score for the item was 0.
If the event had occurred, they identified the degree to
which the event was disturbing, using a five point scale
ranging from 1 = not disturbing to 5 a extremely disturbing.
Thus the range for each event was from 0 to 5. If data were
missing, the mean score of the item was substituted. The
values for the selected events were added to create total
scores for the stress of each period of time (STRS83,
STRS85). The total scores were composed of five types of
stress: economic, time related, children related, familial,
and health related (Appendix B, Q8, pp109-114 and Appendix
C, 019, pp120-123). Economic stress was composed of the
events and stress related to nine items: job loss by major
wage earner, family member's job demotion or retooling,
financial assistance required by own or in-law's family, out
in family income, going into or increasing debt, major
financial loss in family, going on welfare, having to
increase debt due to increased expenses, and forced to dip

into savings.

45

Time related stress was composed of four items: member
accepts time consuming volunteer work, change in member’s
work schedule, outside activities draw adult away from home,
and household chores pile up.

Child related stress was composed of five events
including member demanding new privileges, member dropping
out of school, pregnancy of unmarried member, adult children
having problems achieving independence, and married children
"freezing out" parents.

Familial stress was composed of four events and
accompanying stress levels related to marital separation and
relations with married children, relatives, or in-laws.

Health stress was created from four items: major wage
earner has serious illness or accident; child has serious
illness; grandparents require direct care or become
institutionalized; family member has serious emotional
problems. Data from 1983 created STR883; STRSSS relied on

the 1985 responses.

00' -0 5

' ansentual_nefinitign: The degree of over-all
satisfaction of individuals with their own objective and

subjective financial situation.

Qneratienal_nefinitign: Economic well-being was

measured by four indicators: per capita income (PCI84),

change in family income (CFISS), income adequacy (IAD85),

46
and change in financial situation (CFSBS).

In the 1985 survey, respondents were asked to indicate
which of twenty income brackets best described their 1984
total income before taxes. The income variable was coded
from 1 (1 a less than $5,000) to 20 (20 = $100,000 or more).
To compute per capita income the coded value was divided by
the number of persons in the household. Number of persons
in the household was calculated in this manner: 2 (husband
and wife) plus number of children living at home (Appendix
c, 937, p125 and 011, p119).

For change in family income, respondents were asked in
1985 "In general, which of the following best describes any
changes in your total family income over the past three
years?" Possible responses were: 1 to 6 with 1 2 increase
more than 25%; 2 - increase 5 to 25%; 3 = no change; 4 -
decrease 5 to 25%; 5 - decrease more than 25%, and 6 =
fluctuated up and down over the three years. In the
analysis, these responses were recoded in reverse order so
that 5 = increased more than 25% and 1 - decreased more than
25%. The response for "fluctuated up and down" was treated
as missing for the 5.9 percent of respondents who selected
the ”fluctuated" response. Since the change in family
income variable was measured on an interval scale, the
fluctuation state was not conceptually consistent with the
direction of change being measured (Appendix C, Q40,

p126).

47

Income adequacy measured the extent to which the
respondents thought their income was enough to live on in
1985. Responses were coded 1 = can't buy necessities, 2 =
can meet necessities only, 3 = can afford some of the things
we want but not all we want, 4 = can afford about everything
we want, and 5 = can afford about everything we want and
have some left (Appendix C, Q41, p126).

Change in financial condition was measured by asking
respondents in 1985 to describe any change in their overall
financial condition over the past three years. Responses
were coded 1 = much worse, 2 = worse, 3 = same, 4 = better,

and 5 = much better (Appendix C, Q42, p126).

WW
Conceptual Model

The conceptual model is presented in Figure 2. It

shows the hypothesized relationships among the variables.

48

     

A——=.+

.25.; 33323 ~ 3:2...

 

 

 

__=.+
szT
.351 2::
.215: A \ .2...
5.

32:-

:5

nczhaa

@

3:...

Hypotheses

49

Based on the proposed relationships among the selected

variables, the specific hypotheses to be investigated

relative to middle-aged individuals are presented. The

first set of hypotheses examines the causal effect of the

exogenous variable (STRSS3) on two endogenous variables, as

well as the relationships among all the endogenous,

intervening variables. They are stated as:

Hypothesis

Hypothesis

Hypothesis

Hypothesis

Hypothesis

Hypothesis

Hypothesis

Hypothesis

Hypothesis

1a:

1b:

2:

4:

6a:

Stress (STRSS3) has a negative, direct
impact on perceived health (HLTH83).

Stress (STRS85) has a negative, direct
effect on perceived health (HLTH85).

Perceived health (HLTH83) has a
negative, direct effect on stress
(STRSBS).

Perceived health of past years (HLTH83)
has a positive, direct effect on
present perceived health (HLTH85).

Perceived health (HLTH83) has a
positive, direct effect on economic
well-being (EWBBS).

Economic well-being (EWBBS) has a
positive, direct effect on perceived
health (HLTH85).

Stress (STRSS3) has a negative, direct
impact on quality of family life
(QFL83).

Stress (STRSBS) has a negative, direct
impact on quality of family life
(QFL85).

Quality of family life (QFL83) has a
negative, direct effect on stress
(STRSBS).

50

Hypothesis 8a: Perceived health (HLTH83) has a
positive, direct effect on quality of
family life (QFL83).

Hypothesis 8b: Perceived health (HLTH85) has a
positive, direct effect on quality of
family life (QFL85).

Hypothesis 9: Economic well-being (EWBBS) has a
positive, direct influence on quality
of family life (QFL85).

Hypothesis 10: Quality of family life of past years
(QFL83) has a positive, direct effect
on present quality of family life
(QFL85).

The second set of hypotheses is intended to examine the
effect of four endogenous variables on quality of life.
Hypothesis 11: Perceived health (HLTH85) has a
positive, direct impact on quality of
life (QOL85).

Hypothesis 12: Stress (STRSSS) has a negative, direct
impact on quality of life (QOL85).

Hypothesis 13: Economic well-being (EWB85) has a
positive, direct impact on quality of
life (QOL85).

Hypothesis 14: Quality of family life (QFL85) has a
positive, direct effect on quality of
life (QOL85).

The last hypothesis proposed examines the overall
explanatory power of the structural model for quality of
life, integrating all the variables selected for the present
study. It is stated as follows:

Hypothesis 15: Among middle-aged individuals, the

levels of influence of the independent
variable (STRS83) and the intervening

variables are causally related to the
level of quality of life.

51
W

The research model is presented in Figure 3.
The following equations were developed based on the
structural model in Figure 3.

The structural equations for the latent variable model
are:
“1:711:1IC1
“2=Y21£1+321“1+C2
U3=pain1+pazfl2+ca
fl4gpa1ni+c4
“5:951‘11+953"3+Bscn4+cs
‘16 gpsanz+psan3 Ips¢fl4+pssns+cs
‘17 397393+B74fl4+p7sfls+p7afls *cv

Where £(xi): latent exogenous variables
n(eta): latent endogenous
variables
{(zeta): latent errors in equations
7(gamma): coefficient for latent
exogenous variables
8(beta): coefficient for latent
endogenous variables
X: observed indicators of E
Y: observed indicators of n
6(delta): measurement errors for x

6(epsilon): measurement errors for Y

52

I!

.23.... 3.5.2:; noes—w:

Ssaunuhu

a.» :0 3» 2o

 

 

 

 

 

53
The measurement model corresponding to the proposed
structural model in Figure 3 is presented in Figure 4. The
following equations were developed based on the measurement
model.
The measurement equations are:

x1=111£1+61

x2=12262+62

x3 =13252+53

x, =A‘252+6‘

xs‘fl'szgz'ws

x6 =2'eaaa'wcs

x7 =174E4+67

xegluec'ws

"9‘194£4+69

x1081“, 4541610

x11 'Aiu 5£s+511

x12 3112 0 656 +612

1‘13 =4'13 I 656+613

x14 3114 v 656 +514

X15315! 6564.615

x15"‘3'16 ' 757 +616

x17 =11? I a£a+617

Where X: observed indicators of E

6(delta): measurement errors for x

54

.252 Ens—2:33.. v 2.6:

2» 03 «3 v3 2» N3 So :3

5.: or g. _u .7:

V5—
895 .232. 2:: “a“

335333833

. .‘

23A.
@888

55
W
overview of Analyses

The data in this research were analyzed in two steps.\
First, the measurement model was evaluated. Reliability, ‘
unidimensionality, and discriminant validity were tested.
After the measurement model was evaluated, the second step
was an examination of the structural equation model.

In general, structural equation models are regression
equations with assumptions that allow measurement error in
the explanatory as well as the dependent variables (Bollen,
1989). In these models, assumptions about the distributions
of the observed variables are less restrictive than the
distributional assumptions generally made for ANOVA or
regression analysis. The structural equation models consist
of factor analyses that permit direct and indirect effects
between factors. These include multiple indicators and
latent variables. In brief, these models encompass and
extend regression, econometric models, and factor analysis
procedures.

As implied in the term "linear structural relations",
the model assumes a linear relationship between all
variables, latent and observed variables. Additionally, the
assumptions of homoscedasticity and no autocorrelation for
disturbances or errors across the observations are also made
(Bollen, 1989). The homoscedasticity assumption means that

the variance of zeta (latent errors in the equations) is

56
constant across cases. The no autocorrelation assumption
means that the variance of zeta is uncorrelated with other
zetas in the equation. The assumptions of homoscedasticity
and no autocorrelation do not mean that the disturbances
from two different equations need be unrelated nor that they
need have the same variance.

In structural equation models three components are
presented: (1) the conceptual synthesis of latent variable
and measurement models, (2) path analysis, and (3) general
estimation procedures. Path analysis has three aspects: the
path diagram, the equations relating correlations or
covariances to parameters, and the decomposition of effects
(Wright, 1960).

The conceptual synthesis of latent variable and
measurement models is essential to the structural equation
model. In this regard, factor analysis is utilized to
examine the relation of latent factors to observed
variables. As in econometrics, the structural relation
between observed variables is examined in relation to an
occasional reference to error-in-the-variable situation.

Path analysis provides a means to distinguish direct,
indirect, and total effects of one variable on another. The
direct effects are those not mediated by any other variable;
the indirect effects operate through at least one
intervening variable, and the total effects are the sum of

direct and all indirect effects.

57

The last characteristic of the structural equation
models is general estimation procedures. Joreskog (1973)
proposed a maximum likelihood estimator (based on the
multinormality of the observed variables) for general
structural equation models. This method is the most widely
used estimator today. Generalized least square (GLS)
estimation is another method that offers additional
flexibility in the assumptions. This method can be applied
when arbitrary distributions or elliptical distributions are
found for the observed variables (Joreskog & Goldberger,
1972; Browne, 1982).

In summary, the structural relation technique enables
estimation of both a measurement model and a causal model.
It permits the inclusion of multiple indicators of latent
constructs (unobserved) and the estimation of the
relationships among the latent variables without the
compounding effects of measurement error. In this study the
focus will be on how latent variables or hypothetical
constructs (stress, economic well-being, perceived health,
quality of family life) are causally related to quality of
life. The structural equation model specifies the causal
relationships among the latent variables, describes the
causal effects, and assigns the explained and unexplained
variances. Therefore, the LISREL method estimates the
unknown coefficients of a set of linear structural

equations. The major strength of the LISREL method is to

58
accommodate models that include latent variables and
measurement errors in both dependent and independent
variables, reciprocal causation, simultaneity, and
interdependence.

Since the regional project NC-164 data set is
longitudinal in nature, it is particularly suitable to
causal modeling analyses. With longitudinal data the causal
direction is clear because they reveal a sequence over time.
Although the use of multi-time data cannot remove all
ambiguity about causal directions, multi-time data do allow
confidence in the causality-direction assumptions within a
particular model. In this longitudinal study, auto-
correlation among measurement errors may be assumed since
the same measures were obtained from the same people at two
different times.‘

LISREL VII (Joreskog & Sorbom, 1988), a personal
computer software package, was used to analyze the data.
LISREL VII allows a variety of estimators including

arbitrary distribution estimators.

 

1As in other longitudinal studies, there is a tendency
for the errors in these measures to correlate over time.
When data are gathered repeatedly over time, they may be
correlated due to memory or other retest effects (Joreskog &
Sorbom, 1988, 1989).

59
laasuro validation
W

The validity of the measures was assessed by estimating
the measurement model of all the exogenous and endogenous
variables prior to the simultaneous estimation of the
measurement and structural submodels. This is desirable to
ensure that the alternate indicators of each construct are
acceptably unidimensional (Anderson & Gerbing, 1988).

In this study, STRESS and HEALTH (STR883, STRS85,
HLTH83, HLTHBS) measures were developed by adding the
construct domain to provide formative scales. The
dimensions of the indicators of the latent variables, STRESS
and HEALTH, were not considered to be unidimensional,
reflecting low correlation coefficient among indicators.

The present researcher, therefore, treated these as
formative scales, and summed the scores of each of the
responses in the process of developing the measures. The
remainder of the scales, however, were reflective scales
representing multi-item scales. The unidimensionality of the
reflective scales was examined by separate confirmatory
factor analysis.

Reliability analysis via coefficient alpha and the
tests of unidimensionality through confirmatory factor
analysis via LISREL were conducted prior to the examination
of the structural equation models. According to Anderson

and Gerbing (1988), confirmatory factor analysis is a

60
sufficient condition for an assessment of unidimensionality.
Confirmatory factor analysis jointly assumes internal
consistency and an external measurement model to arrive at
parameter estimates.

In the present study, internal consistency via
coefficient alpha was used to assess the reliability of the
measures of each construct. A coefficient of .60 or above
was considered an indication of the usefulness of the scale
in measuring the variable (Nunnally, 1978).

Test of unidimensionality was considered to be a tool
to achieve convergent validity. xéonvergent validity is
obtained when the correlation between a measure of a
construct and another measure of/the same construct is
significantly different from zero and is sufficiently large
to encourage further examination of validity (Campbell &
Fiske, 1959). In confirmatory factor analysis, convergent
validity is also achieved when the measurement model shows a
satisfactory fit to the data (Bagozzi & Phllips, 1982).

Throughout the study, the covariance matrix, rather
than the correlation matrix was used as recommended by
Joreskog and Sorbom (1988, 1989). When the covariance
matrix is analyzed in LISREL, the normal theory of standard
errors is valid. \Jgsults of confirmatory analysis were
examined to determine whether all loading parameters were
large and loaded significantly on the latent variables

(factors) which they were intended to represent, with small

61

standard errors and significant t-values. The fit indices
of confirmatory factor models were tested in terms offifi}
goodness-of-fit (GFI), adjusted goodness-of-fit (AGFI): and
root mean squares of residuals (RMSR) .2 GFI is an index of
measure of the relative amount of variance and covariance
jointly explained by the model, whereas RHSR is a measure of
the average residual variances and covariances. In addition
to x?, these indicate the overall fit of the model to the
data (Joreskog & Sorbom, 1988, 1989).

If the model poorly fits the data based on the above
criteria, mamresearcher might inspect modification indices.
The modification indices guide possible changes in model

OW

parameters. This decision can only be made when it is

”m... r... _—
\. .“m-u

conceptually plausable.

 

2The goodness-of-fit index is defined as
an a 1 - tr[(E"S - I)2]/tr[(E"S)2 1.
It is interpreted as relative amount of variance and
covariance in s that are predicted by E (Bollen,1989).

The adjusted goodness-of-fit index is the goodness-of-fit
index adjusted for degrees of freedom relative to the number
of variables, and defined as

AGFI = 1 - (p+q)(p+q+1)(l-GFI)/2d,
where, d is the degrees of freedom of the model (Joreskog &
Sorbom, 1939).
Both of CPI and AGFI should be between zero and one,
although negative values are possible for them, meaning the
model fits worse than any other model at all.

Root mean squared residual is defined as

RMSR - [222(8, --0- 2/ (pm) (I>+01+1)J"z
RMSR can be used to compare the fi)t of two different models
for the same data. The GFI can be used to compare the fit
of models for the same or different data.

62

WW

Discriminant validity refers to the degree to which
measures of distinct concepts differ. According to Campbell
and Fiske (1959), discriminant validity is the requirement
that a measure does not correlate too highly with measures
from which it is supposed to differ. That is, measures of
different concepts should share relatively little common
variance. A conventional test of discriminant validity

m ‘w ‘W—

using the structural equations model involves comparing the

._-|.’~

-_.K‘ "7'44v

measurement model to a model in which all the correlation

“fin-.-
.gp‘w‘a" ‘ ,._ \M--.. ...-v. 1*

“___ ”gnu—rm”

coefficients among constructs (i. e., off-diagonal o's) are

v“.Mm-1WH ‘m 7;"

 

constrained to equal one (Bagozzi 8 Phillips, 1982). The
significance of the )8 difference between the two models was
considered to indicate that the constructs are not perfectly
correlated; each construct is uniquely measured. However,
as many others have noted (Bagozzi & Phillips, 1982; Howell,
1987), this is a ”weak" test of discriminant validity in
that it is very unlikely to have all the constructs
perfectly correlated.

In the present study, discriminant validity was tested
by examining each pair of constructs individually (Howell,
1987). Given the model, discriminant validity then was
examined for each pair of the latent constructs by
constraining the relevant correlation to 1 (or any arbitrary
relatively high number i.e., 0.8) and estimating this new

model. Then the-difference in x2 was tested... A significant

63
difference was inferred to indicate that the pair of latent
constructs were not correlated, and it could be concluded
that an adequate level of discriminability had been
achieved. In sum, in this study, discriminant validity was
tested by pairwise comparison of the )8 difference between

freed ¢ and constrained ¢-

Testing the Structural Hod-1

As described earlier, the variables such as HLTH83,
HLTH85, STR883, and STRSBS are formative scales. Although
no reliability measures for these variables were available,
it was assumed that these variables were fallible measures.
As recommended by Joreskog and Sorbom (1988, 1989), the
reliability of 0.85 for one item measures (formative scales)
was used. It was assumed that an arbitrary value of 0.85 is
better than an equally arbitrary value of 1.00.
Consequently, the parameter estimates and standard errors
will be affected by the assumed value of the reliability.
For the error terms, the constraint of the reliability of
0.85 for such variables in LISREL was done by assigning the
fixed value of 0.15 times the variance of each scale to
their respective e, or 9,. For the parameters, Ax and Ay
were fixed at the value of the square root of 0.85
(reliability) times the standard deviation of the scale.

Regarding the reflective scales (e.g., QFL83) one of

the measures (Ay) of each latent construct was constrained

64
to have the value of its standard deviation. The
corresponding error terms (8,) of the reflective scales were
fixed at zero.

The relationships among the variables were examined
simultaneously via analysis of covariance. Maximum
likelihood (ML) estimation, using the covariance matrix as
input data was used to estimate model parameters (Joreskog &
Sorbom, 1988, 1989).3 .According to Anderson and Gerbing
(1988), the ML estimation method is well suited to the test
and development of theory. With large sample sizes, ML
method provides unbiased, consistent, and efficient
estimations (Kmenta, 1971).

In the present study, to check the robustness of the
results, the ML estimates were compared with generalized
least squares (GLS) estimates and ordinary least squares
(OLS) estimates.‘ Bohrnstedt and Carter (1971) found the
OLS estimates were more robust than the ML estimates when
violations of the statistical assumptions, such as normality

of disturbances, were present.

 

3 ML maximizes the likelihood of the parameters, given
the data. It is equivalent to minimizing
F = log IIEII + ms?) - logIISII - <p+q)

‘GLS minimizes
F=1/2tr[ (I - 542)”) ,
equivalent to minimizing the sum of squares of the residuals
weighted by the inverse of the sample covariance matrix
(Joreskog & Sorbom, 1989).

65
M

According to Bagozzi and Yi (1988), non-existence of
any anomalies is a preliminary criteria for model
evaluation. These anomalies are negative error variances
for f, 6, or e; correlations greater than one; and extremely
large parameter estimates.

The overall goodness-of-fit of the present model can be
evaluated with the )8 test. In a structural model, the x2
test is a likelihood ratio statistic for testing a
hypothesized model against the alternative model in which
the covariance matrix is unconstrained.

However, careful inspection should be made because a
valid model might be falsely rejected due to the presence of
a large sample size. Bagozzi and Yi (1988) strongly argue
that the dependence of the )8 test on sample size makes
contrasting outcomes possible. That is, not only might one
falsely reject a valid model because of large sample size,
but also one might falsely accept an invalid model when the
sample size is too small. Because the power of x? is a
function of sample size, the number of estimated parameters,
and the significant level of alpha, it can be increased by
increasing the sample size and the number of parameters
while holding constant the alpha level.

Bagozzi and Yi (1988) suggested that values of AGFI
equal to or greater than about .9 suggest that the model is

meaningful from a pragmatic point of view. It was noted

66

that, in contrast to the )8 test, the AGFI is independent of
the sample size and relatively robust against departure from
the normality assumption (Joreskog & Sorbom, 1988, 1989).

Another way to guide in the evaluation of the overall
model is Bentler and Bonett's incremental fit indices
(Bentler & Bonett, 1980).’ Incremental fit indices also
suggest that the models are generally appropriate when
compared to the null model. The null model is a severely
restricted model hypothesizing complete independence among
the indicators, implying no common factors underlying the
observed variables.“ This can be stated as 2 a Q, with 0
being a diagonal and all off-diagonal covariances are zero.

In the null model, the matrix of Lamdas (A) is an identity

 

sIncremental fit indices provide information about the
adequacy of two comparative models. They are defined as :
p(non-normed fit index) = [(xfi/df.) - (x’_/df_)] /

[ (Xzb/dfb)-1]

A(normed fit index) = [F - ]/Fb
g [X55 - gal/X25

where F. is the fitting function value of a "baseline model"
and F; is the value of the fitting function for the
maintained or hypothesized model (Bentler and Bonett, 1980;
Bollen, 1989). In alternative formula, xfi is the chi-square
estimate for the baseline model and x2. is for the
maintained model. Rho is not normed to lie between zero and
one, whereas delta lies in that interval.

9A null model (Mo) represents ”the severely restricted
model specifying that the variables are mutually independent
(Bentler and Bonett, 1980).” Mo assumes no common factors
among the observed variables. Thus all factor loadings are
set to zero. This can be described as 2 = O, with 0 being a
diagonal matrix. In LISREL, the null model is obtained by
setting LX=DI,FI (at am), TD=DI, FI (at 1-a), PH=DI, FR,
with known variance. This would give the same results as
the model of LX=DI, PH=DI, FR, TD=FU, FI and that of
LXIFU, FI, TDIDI, FR.

67

matrix and the errors of the lamdas (6,) are zero.

According to Bentler and Bonett (1980), even if one has
a large or small sample, values of delta greater than .9 can
be considered an adequate fit. Further, the model may be a
substantial statistical improvement over the null model.
The nonnormed fit indices (p) and normed fit indices (A) of
the hypothesized structural model indicate the degree of

adequacy of two comparative constructs.

68

CHAPTER IV

RESULTS '9

u e 0

Reliability and Unidimensionality

In this study, as stated in Chapter III, STRESS and
HEALTH (STRSBB, STRSBS, HLTH83, HLTH85) measures were
treated as one item scales. Therefore, the dimensions of
the indicators STRESS and HEALTH were not tested. The
unidimensionality of the remainder of the scales, however,
was tested by separate confirmatory factor analysis.7

Table 1 summarizes key parameter estimates, and indices
of the confirmatory factor models and the scale reliability
scores. With regard to reliability, all constructs showed(
levels of coefficient alpha well above the conventionally.
accepted criterion of .60. Among the measures, EWBBS had
the relatively-lowest reliability scores (.676). Some
degree of fluctuations in the responses was implied by the
results because reliability refers to consistency of

measurement (Bollen, 1989). However, the remaining measures

showed sufficiently large scores (above .80), indicating the

 

7Throughout the study, the covariance matrix was used
in the analyses. Descriptive statistics including
correlation coefficients, means, and standard deviations are
presented in Appendix A.

69

Table 1 Measurement Models: Key Parameter Estimates, t-values, Model Fits, and

 

 

 

Scale Reliability

Construct! ML Fit Scale
1m Estimates t-values Indiees Relibility
QFL83I 28(3) = 443.48

SFL83 _ 1.179" . -- p = .000 .869

SSP83 . 1.167 40.742 GFI = .871

SCH83 .923 39.187 AGFI = .571

SCF.83 .858 30.020 RMSR=.129
QFL85/ {(3) = 331.58

SFL85 .952h -- p = .000 .821

SSP85 .923 36.361 on = .905

SCH85 .584 27.538 AGFI = .683

801385 .507 20.543 RMSR - .088
was 18(3) . 177.51

PC184 .560 14.935 p = .000

cms .535 25.104 0171 = .944 .694

IAD85 .488 23.663 AGFI - .812

CFsss .688" -- RMSR - .078
001.85

001.1 .856 35.262 1800 = 400.65

001.2 .885 40.919 p = .000 .943

001.3 .852 38.697 GFI = .942

001.4 .606 28.414 AGFI = .903

001.5 .920 41.001 RMSR = .028

001.6 .977 40.931

00L7 1.028 43.486

001.8 .816 37.828

00L9 .746 32.000

”mm were not tested.

70
measures of the respective constructs are internally
consistent. The results of confirmatory analysis showed

that all loading parameters were large and loaded

up...
\-.....‘....- M-

significantly on the latent variables which they were

-V—uh ~h~‘l

intended to represent with small standard errors and
w”/ "e.,- '~- v.1.»awm a...” w.» .w J __H _ .1 .._

significant t-values. The results prov1ded evidence of

‘wv-I' ,2,-1\M.fl-.~ 1“,.

\

[taut-.1.
s. Wflmw wa— ‘. u

un1d1mens1onal1ty and convergent validity. Thus, for
example, the four indicators of QFL83 (SFL83, SSP83, SCH83,
and SCE83) were all acceptable for quality of family life
measure.

The fit indices of Table 1 indicate that none of the
confirmatory factor models were accepted by the )8 tests.
However, when goodness-of-fit (GFI) indices and root mean
squares of residuals (RMSR) were examined, the lack of
significance in )8 did not appear to be a serious problem
except for QFL measures (QFL83: x2 =- 443.3, d.f.== 3, p
=.000, GFI = .871, AGFI = .571, RMSR = .12; QFL85: x2==
331.6, d.f.= 3, p = .000, GFI = .905, AGFI = .683, RMSR =
.088).

To improve the model fit, the researcher inspected the
modification indices. As a result, it was decided to free
out the parameters between SCH83 and SCE83, and between
SCH85 and SCE85. This was done because of the possible
existence of correlated errors of the measures between SCH
(Degree of satisfaction with your relationship with your
children) and SCE (Degree of satisfaction with the

E. W A :
@fivééééég

[/ALNMMMVQ
[III-ll

72
relationship your children have with each other) for 1983
and 1985 data. In other words, the two indicators might be
conceptually related. Therefore, assuming uncorrelated
error variances among the indicators (SCH83 and SCE83, SCH85
and SCE85), the Goodness of Fit Index indicated the models
poorly fit the data. Examination of the measurement models,
after allowing for correlated errors, provided better fit
for QFL83 and QFL85 (Table 2). For example, x? for QFL83
decreased from 443.3 (d.f.= 3) to 23.05 (d.f.= 2), and GFI
improved from .871 to .993, in Table 2.

The test of the overall measurement model was conducted
using confirmatory factor analysis. The results appear in
Figure 5. Although GFI seems reasonably adequate along with
RMSR, the x’ seemed to be large (xzm .. 887.1, GFI = .933,
AGFI - .900, RMSR = .143). Acknowledging the
characteristics of the longitudinal data set, further, auto-
correlations among error variances between the measures
QFL83 and QFL85 were allowed (Figure 6). After allowing for
auto-correlation between the indicators, the model showed
considerable improvement. The results of the overall model
fit with the modification is indicated in Figure 6. The
results of the overall model fit showed improvement in terms
of x’. GFI, AGFI, and RMSR (x29. -- 486.3, GFI .962, AGFI =
.940, RMSR =- .141). The x2 difference test results led to
the conclusion that the Measurement Model II in Figure 6 was

more adequate than Model I (dx’(4) - 400.8, p < .001) .

73

Table 2 Measurement Models After Modification: Key Parameter Estimates, t-values,
Model Fits, and Scale Reliability

 

 

 

Construct! ML Fit Scale
Items' Estimates t-values Indiees Relibility
QFL83/ {(2) = 23.05
SFLB3 1.158" -— p = .000 .869
SSP83 1.075 34.012 GFI == .993
SCH83 .815 31.126 AGFI == .963
SCE83 .742 24.387 RMSR = .116
QFL85] {(2) = 28.08
SFL85 .934b -- p = .000 .821
SSP85 .875 31.559 GFI = .991
SCH85 .538 24.024 AGFI =- .956
SCESS .459 18.114 RMSR s .049
EWB85 {(3) = 177.51
PCI84 .560 14.935 p = .000
CF185 .535 25.104 GFI 2: .944 .694
IAD85 .488 23.663 AGFI - .812
CFSBS .688h -- RMSR - .078
Q0185
QOLl .856 35.262 {(27) - 400.65
QOLZ .885 40.919 p = .000 .943
QOL3 .852 38.697 GFI == .942
QOL4 .606 28.414 AGFI - .903
QOLS .920 41.001 RMSR = .028
00136 .977 40.931
QOL7 1.028 43.486
QOL8 .816 37.828
QOL9 .746 32.(X)0
WWW not tested.

b:Constrainedpanmete1-8 to 1.0.

74

2: Boos. €050.38: m 059“.

we... I Ems—m
ova. I E0<
New. I E0
am I 5

n63. u N x

xxx 3. I. . ans

2» 3» 3» 3o n3 NS 2» So 3 3 S 3 no 3 an «a 5

E EEEEEEEE—WEEEEEBE

as: he: 6.. a. _ _.= . .. m: ”.5
z: ‘5
96869809

 

 

 

75
Discriminant validity

In this study, discriminant validity was tested by
examining each pair of constructs individually. Given the
model, discriminant validity then was examined for each pair
of the latent constructs by constraining the relevant
correlation to 1 (or any arbitrary relatively high number
i.e., 0.8) and estimating this new model.8 IDiscriminant
validity was tested by pairwise comparison of the x2
difference between freed ¢ and constrained ¢~ The
correlation matrix of the latent variables is shown in
Table 3.

The results of the )8 tests using confirmatory factor
analysis are shown in Table 4. An example of the procedure,
for testing discriminant validity between £6 and £7 (QFL85
and QOL85), one of the most stringent one in the pairs, as
follows. The new )8 score after constraining 4475 to 1.0 was
presented in Table 4 (1766.4) . The difference in x2 was
obtained by subtracting the original x? value (498.14,
presented in Table 3) from new'x'2 (1766.4). As a result,
the difference in )8 yielded dx2(1) = 1268.3 (p < .005); and
Ho: ¢ = 1.0 was rejected. Thus it was concluded that the
two measures, QFL85 and QOL85, are uniquely developed,
measuring different concepts. Further tests of x2

difference for discriminability indicated that all of the

 

'In testing discriminant validity, the correlation
matrix was used.

76

Table 3 Correlations among Constructs (<I> matrix)

 

 

£1 £2 £3 £4 £5 £6 £7 £8
£1 1.000
$2 -.003 1.000
53 .014 -.118 1.000
54 .053 .033 -.067 1.000
.55 .259 .017 .007 .053 1.000
£6 -.021 .436 -.127 .070 .051 1.000
57 .027 .272 -.091 .252 .103 .680 1.000
£8 -.051 -.275 .137 -.052 -.049 -.106 -.057 1.000

 

x2 = 498.13
d.f. = 98

p = .000

GFI = .961
AGFI = .939
RMSR = .046

.ndH 8 Bee 803 ”.0 no
sea a 85.. 26; re 8
5.3" an 3.8 803 «.0 8

.c be 3.8 203 ”.0 B
6.3“ 3 none 0.63 90 ”a

77

 

 

.1 .882 .36: 8.88 .568 6.88 .53“ 6.88 we
8.8: .36: 58$ .38,” 6.38 .38 8
i .28 8.28 8.2.: .08“ .38 3
1. 6.2.8 .988 .38 .384 R
.1 .38" .38 .988 z
...- 8.88 .688 m
..l afiwmv Nw
i E
R E 3 a 8 E m m

 

82; 1:88:35 .6 8.8. 4 use.

78
measures in the present study were different from each
other, and it was concluded that an adequate level of
discriminant validity among the constructs was achieved in
the measurement model.

From the overall evidence on internal consistency,
factor structure, and reliability, with modification of the
original measurement model, the researcher concluded that
adequate levels of the measurements were achieved. Thus all

the constructs were utilized in the model to be tested.

Iesting_fhe_§tructural_ncdel
Figure 7 presents a revised model of the original
structural equation model (Figure 3) adjusted for the
results of measure validations. In the revised model,
correlations and autocorrelations of the error terms were
allowed due to the characteristics of the longitudinal data
set. The results of LISREL analysis on the revised model

are presented in Table 5.

Modal Fit:

The results were examined for anomalies as described in
Chapter III, and none were found. Table 5 contains the
global measures of fit of the structural model as well as
the parameter estimates. The overall effect model of the ML
and GLS methods failed to meet the )8 test (ML: x210, = 482.1,

p=.000; GLS: xflm = 389.8, p = .000). Thus, the hypothesized

79

 

32322—32 5.3. .0921 .053-9:...“ h 9:3:

Sean» e an

.2. EEEE

2» 3» fl» an»

3 a 5

EEEE @ BEBE
o. 5
eh.‘.—

. .. s
2» en. 2 m I 5 _e

 

 

 

 

80
model was rejected by the )8 test, and the model failed to
fit in an absolute sense. However, careful inspection was
made because, as discussed in Chapter III, a valid model
might be falsely rejected due to the presence of a large
sample size, a characteristic of this study.

All of the values of GFI (.962 for ML; .970 for GLS)
and AGFI (.943 for ML; .954 for GLS) are greater than .9. A
general rule-of-thumb is to accept the hypothesized model if
the AGFI index is greater than .9. Recall that Bagozzi and
Yi (1988) suggested that values of AGFI equal to or greater
than about .9 suggest that the model is meaningful from a
pragmatic point of view.

Bentler and Bonett's incremental fit indices are shown
in Table 5. The null model is the severely restricted model
hypothesizing complete independence among indicators.
Fitting the null model gives xfl“ - 9919.5 (ML), 2742.4
(GL8); p - .ooo; GFI - .491(ML), .786(GLS); AGFI = .427
(ML), .759 (GL8); RMSR 8 5.129 (ML), 6.449 (GLS).

According to Bentler and Bonett (1980), the nonnormed
fit indices (p) and normed fit indices (A) of the
hypothesized structural model indicated the degree of
adequacy of this model compared to the null model. Whether
one has a large or small sample, values of delta greater
than .9 can be considered an adequate fit. The results of
the present study (A a .951 (ML); p a .935 (ML)) indicated

that the model was a substantial improvement over the null

Table 5 Structural Model: Parameter Estimates, t-values, and Model Fits

81

 

 

ML 61.8 01.8

Parameters estimates t-values estimates t-vnlues estimates t-values
7,. -.052 4.712 -.054 4.808 -.353 4.658
12, -.356 -10.763 -.351 -10.742 -.548 -9.439
6,. -.023 -.680 -.021 -.641 -.008 4.199
8,, .013 .441 .014 .474 .001 .686
3,, .036 1.702 .036 1.761 .002 1 . 129
3,, .257 8.663 .263 8.791 .259 8.767
B” -. 103 -4.583 -. 104 4.496 -.011 -3.929
6., .340 16.965 .335 16.212 .322 18.587
5,, .005 . 160 .011 .373 .130 . 138
as -.072 ~2. 852 -.075 ~2. 886 -.254 -1.629
3,, .002 .086 -.016 -.712 -.118 -1.701
5,, .057 1.268 .074 1.579 .441 .819
3.. .064 1.684 .079 2.015 -.001 -.015
B1, .291 8.680 .325 9.331 .383 9.757
6,, .039 1.564 .040 1.585 p .007 1.642
3,, .057 2.676 .056 2.637 .006 3.026
6,. .706 29.231 .696 28.073 .318 30.665
Am 1441? -— 14.414' _.
Am 1.000” -—- 1.000b -—
1.,” .934 38.601 .948 38.029
x, .704 34.345 .705 34.048
1,, .632 26.890 .626 26.255
>50 .450' _— .4so- ..
A”, . 865 14.630 .952 15.338
x,“ .790 21.250 .801 22.030
g. .748 20.985 .817 21.866

 

Table 5 (cont’d)

82

 

 

ML GL8 0L8
Parameters estimates t-values estimaws t-values estimates t~values

1,“... 1.000“ -- 1.000b --

79..., 16.934‘ -- 16.934' --

1,,” 1.000 —- 1.000 .—

g... .967 38.296 .971 37.213
N,“ .608 27.883 .612 27.246
)5,“ .496 19.660 .496 19.401
N167 1.664' -- 1.664' .—

2," 1.745' -— 1.745'I --

it" .997 23.314 .984 23.121
142, 1.388 24.191 1.337 23.602
#1” .981 23.240 .937 22.629
4,, .455 16.418 .425 16.186
#7,, .931 22.831 .910 22.509
4.. .705 21.678 .683 21.248
$1, .489 22.353 .453 21.737
9," 36.658' -- 36.658‘ -

9a .006 .237 .025 1.086
9.3 .864 22.717 .8“) 21.053
9... .672 24.937 .613 23.093
9.5, 1.604 27.064 .976 24.935
6,“ .036' - .036' --

9m 1.541 25.929 1.277 23.603
9,. .377 21.447 .339 19.110
9,, .363 21.924 .272 18.196
9...,” .185 10.389 .185 11.132
9,"... 50.597' -- 50.597' --

9.12.12 .089 5.806 .092 5.992

 

83

Table 5 (cont’d)

 

 

ML GLS
Parameters estimates t-values estimates t-values
9.13.13 .515 21.952 .482 20.414
9,“... .500 26.060 .453 24.337
9.15.13 .767 27.133 .674 24.906
9“,. .161' -— . 161‘ -
9,,” -.068 -6.610 -.059 -5.702
9,.” .252 12.205 .224 10.837
9,,“ .107 8.303 .096 7.709
6,.” .203 10.153 .176 9.113
9.5. .406 15.794 .354 14.061
9.15.14 .242 13.800 .199 12.038
9," .538' -- .538‘ --
x’ 482.1 389.8
d.f. 102 102
p .000 .000
GFI .962 .970
AGFI .943 ' .954
RMSR .188 .517
A .951 ' .858
p .948 .853

 

a: A’s and 0’s were fixed at 01'” a and (l-a)a’, respectively.
b: Constrained parameters to 1.0.

84
model, and it was concluded the model is adequate in terms
of the relative fit. Therefore, despite the rejection of
the model based on the )8 test, Bentler and Bonett's normed
fit index indicated that the model accounted for a
significant proportion of information, from a pragmatic
point of view. Further, any achievement in increasing the
fit would be trivial. Overall, the global measures of fit
indicated the hypothesized model had overall adequacy and

satisfactory fit.

Hypotheaaa Teat

A large number of hypotheses based on theoretical
grounds and previous empirical findings were developed in
Chapter III. The structural model in Figure 3 incorporates
these hypotheses. The following section will discuss the
test of these hypotheses. Table 5 reports the key parameter
values obtained for testing hypotheses by estimating the
revised model of Figure 7 after the validation of
measurement. The discussion will be based mainly on the ML

results, supplemented by GL8 and OLS results.

Hypothesis 1a: Stress (STRSBB) has a negative, direct
impact on perceived health (HLTHB3).

Hypothesis 1b: Stress (STRSBS) has a negative, direct
effect on perceived health (HLTHBS).

The hypotheses predicted stress to have a negative

direct effect on health. For the 1983 data, the

85
hypothesized effect of stress on perceived health was
supported (7" = .52, t = -1.72, p < .10). However,
hypothesis 1b was rejected (B,3 - .005, t = .160). The
results indicated that the impact of stress on perceived

health was marginally significant in 1983.

Hypothesis 2: Perceived health (HLTH83) has a
negative, direct effect on stress
(STRSBS).
The results indicated the effect of perceived health on
stress at a later time was insignificant (B31 = .013,

t = .441). Thus, the hypothesis was rejected.

Hypothesis 3: Perceived health of past years (HLTH83)
has a positive, direct effect on present
perceived health (HLTH85).

The hypothesis was supported (5” - .260, t = 8.66,

p < .001). The results were consistent for all three
estimation methods: ML, GL8, and 0L8. The results indicated
that middle aged individuals with poor perceived health in

1983 also perceived their health to be poor in 1985.

Hypothesis 4: Perceived health (HLTH83) has a
positive, direct effect on economic well-
being (EWBBS).

The posited effect was marginally significant
(34 8.036, t - 1.702, p < .10). Thus perceiving health to
be good in 1983 contributed to a higher level of economic

:.l‘.- ,

well-being in 1985..

86
Hypothesis 5: Economic well-being (EWB85) has a
positive, direct effect on perceived
health (HLTH85).
Hypothesis 5 examined the effect of economic well-being
on perceived health. It was not supported. The coefficient

for the hypothesis was insignificant (5“ = .057, t = 1.268).

The results were consistent for all three methods.

Hypothesis 6a: Stress (STRS83) has a negative, direct
impact on quality of family life
(QFL83).
Hypothesis 6b: Stress (STRSBS) has a negative, direct
impact on quality of family life
(QFL85).
The hypothesized effect of stress on quality of family
life (QFL) was significant (121 8 -.356, t = - 10.763,
p < .001, B“ = -.072, t = -2.852, p < .01). The results
indicated that stress in 1983 had a negative impact on QFL
in 1983. The results were similar for 1985. Consistent
results were observed by all three different estimation
methods. The finding supported the conclusion that among
middle aged individuals, quality of family life at any one

time was affected by the stress from recent events.

Hypothesis 7: Quality of family life (QFL83) has a
negative, direct effect on stress
(STRSBS).
The reverse effect of quality of family life on stress
was significant (B32 = -.103, t = -4.583, p < .001). Thus,

hypothesis 7 was supported. -It was concluded that as family

87
life deteriorated, individuals felt more stress; likewise,
as the quality of family life increased, stress levels

decreased.

Hypothesis 8a: Perceived health (HLTH83) has a
positive, direct effect on quality of
family life (QFL83).

Hypothesis 8b: Perceived health (HLTH85) has a
positive, direct effect on quality of
family life (QFL85).

The posited effect of perceived health on quality of

family life was rejected in both years (B31 2 .013,t - .441,

and B“ - .039, t - 1.564).

Hypothesis 9: Economic well-being (EW385) has a
positive, direct influence on quality of
family life (QFL85).

The results indicated the hypothesis was significant

(Bu - 0.064, t - 1.684, p < .10). Economic well-being had

direct positive influence on quality of family life.

Hypothesis 10: Quality of family life of past years
(QFL83) has a positive, direct effect on
present quality of family life (QFL85).

The hypothesis was supported (3n = .340, t = 16.965,

p < .001). The results indicated that quality of family
life in 1985 was influenced by the quality of life at an

earlier time.

88
Hypothesis 11: Perceived health (HLTH85) has a
positive, direct impact on quality of
life (QOL85).
The results indicated that perceived health was
significantly related to the individual's QOL. The

coefficients for the effects of perceived health were

significant at p < .05 level (3” = .057, t = 2.676).

Hypothesis 12: Stress (STRSBS) has a negative, direct
impact on quality of life (QOL85).

The results indicated that the direct effect of stress
on quality of life was insignificant (Bn 3 .002, t a .086).

Thus hypothesis 12 was rejected.

Hypothesis 13: Economic well-being (EWB85) has a
positive, direct impact on quality of
life (QOL85).

The findings supported the hypothesis (Bu = .291,

t = 8.680, p'< . 001). It was concluded that greater
economic well-being resulted in higher levels of quality of

life.

Hypothesis 14: Quality of family life (QFL85) has a
positive, direct effect on quality of
life (QOL85).

The effect was significant (876 = .706, t = 29.231,

p < .001), and thus it was concluded that quality of family
life was a strong predictor of quality of life.

In sum, the results indicated that people’s quality of

89
life was affected by their perception of health, their
economic well-being, and their quality of family life.
Stress did not directly influence individuals' evaluations

of quality of life.

Hypothesis 15: Among middle-age individuals, the
levels of influence of the independent
variable (STRS83) and the intervening
variables are causally related to the
level of quality of life.
Hypothesis 15 examined the overall structural model.
As described earlier in this chapter, although the overall
model was rejected based on the )8 test, other indices
strongly supported an adequate level of fit. Thus, it was
concluded that the hypothesized model adequately represented

the causal relationships among the latent constructs.

W1
Figure 8 contains the parsimonious model; that is, the

refined model which does not contain paths that are
statistically insignificant. In the present study, the
parsimonious model was obtained by constraining all the
insignificant paths, t-values less than i 1.645 (p = 0.10),
to be (zero. The results of the )8 difference test showed
that the difference between the hypothesized model (Figure
7) and the parsimonious model (Figure 8) is not significant.
The statistical comparison between the hypothesized model

90

and the parsimonious model is given in Table 6.

Table 6 )8 Difference Test for Parsimonious Model

 

 

Models x2 df p GFI AGFI RMSR
Overall effect model 482.1 102 .000 .962 .943 .188
Parsimonious model 486.8 108 .000 .962 .946 .245
)8 Difference dx2 = 4.7 ddf = 6 p > .05

 

Based on the results of )8 difference test and the
almost identical results of GFI, AGFI, and RMSR, it was
concluded that the parsimonious model, as a whole,
represented the causal relationship among the constructs as
well as the hypothesized model. The parsimonious model
would be more desirable because of its relative simplicity.

91

new. .. :22:
o3. . m9
«8. u to
So. .. a

so. .. a
$3 .1. «a

  

.322 395563.”. a 2:98.

.2. con. man. N.

n8.
4 EBB—u ..
8»... v3.68. eases-.8...
9.. 4. Vans.

BEBE % .. BEBE

.. .2. Wk - we... ._

  

Chapter V

SUMMARY, DISCUSSION, AND IMPLICATIONS

The main purpose of this research was to develop a
model of quality of life and to empirically test the
relationships between selected dimensions of quality of
life. The dimensions of quality of life tested were quality
of family life, perceived health, economic well-being, and
stress. Quality of life, in this study, was based on the
conceptualization of Campbell and his associates (1976). In
contrast to numerous studies on quality of life and life
satisfaction, not much has been known about the quality of
life of middle aged people, the focus of this study.

The theoretical perspective of the present study was a
family ecological approach, examining humans in their
environments. The primary focus was on the effects of
environmental dimensions, particularly those related to
family life, on the evaluation of an individual's quality of
life as an outcome of interaction between individuals and
their environments.

The findings of this study were based on the data
collected for the regional research project, NC-164. The
data were obtained from two different periods of time, 1983

and 1985. The longitudinal data set was particularly

92

93
helpful in hypothesizing and testing the direction of
causality among related concepts. It was considered
especially useful for capturing the dynamic processes and
the cumulative effects of causal indicators on quality of
life.

Hypotheses were developed based on theoretical and
empirical grounds. Prior to the testing of these
hypotheses, the measures of the constructs were examined for
adequacy. Internal consistency via coefficient alpha was
used to test the reliability of each scale and confirmatory
factor analysis was used to examine the unidimensionality of
each scale. Discriminant validity was tested via a series
of pairwise comparisons using confirmatory factor analysis.
These tests screened the adequacy of each measure.
Reliability scores of all of the measures were greater than
the conventionally accepted criterion. The results of
confirmatory factor analysis assured unidimensionality of
the measures. An adequate level of discriminant validity
was found.

Tests of hypotheses were based on the hypothesized
structural model using LISREL. Hypotheses were tested via
significance and direction of individual parameter
estimates. Maximum likelihood estimates were compared with
generalized least square estimates along with ordinary least
square estimates to check the robustness of parameter

estimates.

 

 

ef’
pa:

in

was
qua
res
whi
of '
(Am
197E

81.

life
Subj.
qual;
(Aka:

bein

94

The present study investigated the effects of demands
and resources of the environment on the satisfaction of
family life and quality of life. It was found that
individual evaluations of quality of life are influenced by
past experiences. Longitudinal data made possible the
findings which are useful in the explanation of cause and
effect relationships. In this study time, identified as
part of the natural environment, is an important component
in predicting quality of life.

‘The effect of quality of family life on quality of life
was strongly supported. It was the strongest predictor of
quality of life of the middle aged individuals. This
research confirms many other studies about quality of life
which have found that satisfaction with family life is one
of the highest predictors of overall quality of life
(Andrews & Withey, 1976; Campbell, 1981; Campbell et al.,
1976; Bubolz et al., 1980; Sontag et al., 1979; Walker et
al., 1990).

Another important dimension in predicting quality of
life was economic well-being. Measured by objective and
subjective components, it had a significant impact on
quality of life. The finding confirmed other research
(Akerman & Paolucci, 1983).

Of special interest was the effect of economic well-

being on the quality of family life. It was found to be

95
rather weak. This weak effect may be due to the
characteristics of the sample population. The middle years
of life are characterized as a time when children are
becoming independent and launched, and economic resources
are relatively high due to less demand compared to the other
stages of the life cycle.

Perceived health was a dimension which had a
significant impact on quality of life. This is consistent
with previous research (Campbell, 1981; Campbell et al.,
1976; Edwards & Klemmack, 1973; Near et al., 1978). Because
personal health problems may increase as people get older,
health concerns may be especially salient to those in their
middle years of life. Stress had a marginally negative
effect on their perceived health. Therefore, it can be
concluded that stress had a negative indirect effect on
quality of life through perceived health among the middle
aged.

Perceived health, measured by health symptoms, had an
indirect effect on quality of life through economic well-
being. The results are consistent with previous findings
(Kratzer, C., 1991). The effect of perceived health on
economic well-being may be attributed to the interference of
health problems with the ability to work and, thus, to have
financial needs met. This finding supports the family
resource management model, which identified health as one

kind of family resource. In this study, having more

96
resources in terms of good health would bring more economic
or financial satisfaction, and result in a higher quality of
family and quality of life.

A direct effect of stress on quality of life was not
found. However stress had a direct, negative impact on
quality of family life. Higher stress levels reduced the
quality of family life. Based on family ecological theory,
reciprocal interactions between stress and quality of family
life were hypothesized and examined. For example, the
effect of stress on quality of family life was hypothesized
to be negative, and this effect would further negatively
affect stress. The results of this study support the
interactive causations of stress and quality of family life.
That is, stress (STRSS3) negatively affected quality of
family life (QFL83) which, in turn, affected degree of
stress at a later time (STRS85). As quality of family life
in 1983 decreased, stress levels rose in 1985.

Although previous research found that stressed people
were more likely to have illness, depression, anxiety, low
self-confidence (Caplan et al., 1980; Hahn et al., 1964;
Rabkin & Streunig, 1976), the present study did not support
the previous findings. One possible explanation might be
related to the measure of the latent construct, perceived
health. In the present study, perceived health was
operationalized by asking twelve items of general

difficulties which can be an indicator of illness. The

97
respondents answered with the frequencies of the symptoms,
most of which were physical rather than psychological, and
quite different from those used in the other studies. The
results of the present study seem inconsistent with the
previous findings.

This study has demonstrated the usefulness of using a
family ecological model along with a family resource
management model to study quality of life. From an
ecological perspective, the evaluation of quality of life
can be assessed as an outcome of functioning of the family
ecosystem over time, based on satisfaction with various
dimensions of the environment.

It is evident that family life affects quality of life.
Economic well-being measured in objective as well as
subjective terms is another important predictor variable.
Perceived health, a measure of human resources, affected
people's evaluation of quality of life. Stress, being an
outcome, or product of interactions between the organism
(individual) and the environment (family, other family
members, or work), indirectly influenced quality of life.
The effect was mediated through quality of family life.
This result supported other research which hypothesized that
stress declines with age (Herzog & Rodgers, 1986). It has
been argued that stress may be created by ”daily hassles"
that people experience when carrying out routine

transactions with the environment (Manner, Coyne, Schafer, &

98
Lazarus, 1981). For individuals in their middle years,
these daily hassles decline relative to those in earlier
stage in life cycle, as they become released from many of
the family roles and constraints that make life difficult.
Thus, the effect of stress on quality of life may not be as
great for those people in middle years compared to their
younger years.

From the family resource management model, perception
of quality of life can be considered as an assessment of the
extent to which demands are met with available resources
through processes of planning and implementing. This study
focused on the effect of inputs (demands, goals, and
resources) on quality of life as an output. The findings of
this study examined four major dimensions, representing
demands and resources, which affect quality of life. These
are stress, health, economic well-being, and quality of
family life.

Relationships with one's spouse and children and other
members outside the nuclear family can be a source of
stress, as well as an important resource which can play an
important role in evaluating quality of family life. Stress
related to economic environment, including work environment,
and time demands in family life were included in the stress
variable. Perceived health, as an indicator of level of
health, was an important contributor to quality of life.

Good health enables one to attain and manage resources

99
whereas poor health imposes additional demands upon
resources. Economic well-being represents the assessment of
overall availability of economic resources for meeting needs
and achieving goals.

In doing a secondary analysis, the present researcher
acknowledged that this study has mainly two limitations.

The first limitation related to the sample of the original
data set. The original study, Regional project NC-164, drew
its sample from the population of middle years individuals.
Particularly, the data were collected from husbands and
wives between 35-65 years of age with at least one child,
using a probability sampling method. The original sample
was gathered with more emphasis on rural populations. Thus,
the findings of the present study can only be generalized to
those individuals in their middle years of life cycle.

The second limitation relates to the issue of
measurement. The perceived health variable was
operationalized by creating a scale from questions related
to frequency of various symptoms or difficulties. These
questions hardly captured the objective and subjective
levels of health which the present researcher

conceptualized.

100
We
Implications for Future Raaaareh

Efforts to improve quality of life must focus on ways
to enhance quality of family life, with particular attention
to the dimensions which have been identified as making
significant contributions to quality of life. Endeavors to
improve quality of life must consider the family as an
ecosystem in which inputs of demands and resources from the
environment impact on the family and in which family actions
and conditions feedback to the environment.

The results of the present study have implications for
future research in this area. Continuous work with quality
of life in identifying influential factors is imperative.
There might be other.dimensions which may influence
individuals' evaluation of quality of life. A holistic
approach developing a predictable model for quality of life
to include such dimensions as community environment,
housing, work satisfaction, and leisure, is strongly
recommended. It is also strongly recommended that research
efforts include time dimension via longitudinal data sets.

Examination of the usefulness of the model of the
present study with other populations is desirable. There
might be some differences in structural relationships in
other subgroups. For instance, quality of life of those in
their early stages of life cycle might be determined by

different influential dimensions. Structural differences

101
between men and women, or rural and urban groups might be
revealed.

In today's world, where countries do not exit in
isolation, cross-cultural studies of quality of life are
worthwhile. People in different cultures have different
value systems forming their social cultural norms. It can
not be denied that compared to the United States, Korea, the
home country of this researcher, has substantial, historical
differences in family and social structures. Although Korea
has undergone rapid social and economic changes, there may
be pertinent characteristics that distinguished it from
other culture. A comparative study of quality of life in
Korea would provide an opportunity to investigate the

universality of the model.

Implications for Family Ecological Theory

Utilizing a family ecological framework, the study was
designed to determine quality of life over time in terms of
cause and effect relationships. The findings from this
study suggest some important implications for family
ecological theory.

The results of this study indicated that one of major
considerations in doing research on quality of life from an
ecological perspective is the inclusion of the time
dimension. This study found that what happened in the past

had a causal influence on well-being at a later time. For

102
instance, the research points out that stress from total
environments at one point in time had influences on quality
of family life, which, in turn, affected stress in the later
year. Since humans and families exist in and over time, and
across periods of time through generations (Morgan, 1985),
and are dynamic and ever changing, the inclusion of the time
dimension will provide an opportunity to explore the
family's movement and change through time. Thus, for the
sake of family ecological theory development it is important
to more clearly enunciate the time dimension. This study
illustrated that there are both change and stability in the
family ecosystem.

Another consideration for the researcher oriented with
ecological theory is to conceptualize quality of life as an
outcome of the family system and multiple indicators,
recognizing that quality of life encompasses several
interrelated dimensions. It was originally assumed, in the
research of quality of life by Campbell and his associates
(1976), that people's experiences and behavior derive from
interactions with their environment. The environments were
classified as various domains such as family, job, health,
and so forth. The findings of this study confirm the
dependence of quality of life on such domains of interest as
family life and health. Inclusion of more domains or
environments in an attempt for a holistic approach is

strongly recommended for ecological theory development.

103

Finally, qualitative research is profitable in addition
to quantitative research in terms of ecological theory
development.' Qualitative research, such as indepth
-interviews, may have theoretical value in investigating the
standards individuals use to evaluate quality of life and
the relative importance of various dimensions as well as the
evaluation process. Both types of research are necessary to
inform policy makers about the dimensions and conditions of

the ecosystem which are most valued.

Implications for Practice

According to Hunter and Sundel (1989), the major gap in
practitioner knowledge about the middle years has been the
lack of research. The availability of empirically validated
knowledge of middle years could be useful to help
professionals concerned with middle age adulthood.
Therefore, the results of this present study have important
implications for practitioners who seek to help people
improve their sense of well-being. The findings of the
present research showed what were the strong predictors of
quality of life.

Among them satisfaction with family life was the
strongest predictor for quality of life. Professionals can
help middle aged individuals to have a higher level of
quality of family life. They could help couples to

communicate and spend time together around issues of

104
importance to each individual to bring higher marital
satisfaction. The middle years of the life cycle are often
characterized by role overload, due to dual role
expectations from children and aging parents. Professionals
and counselors might help families in their middle years to
get aid from support systems, such as relatives, friends,
neighbors, and public or private social services (Hunter 8
Sundel, 1989).

Stress was shown as an important predictor of quality
of life. Although the effect was not direct, it had a
strong influence through quality of family life. Counselors
and therapists can help midlife individuals to implement
stress reduction techniques. Professionals can also help
middle aged people evaluate their resources to cope with the
stresses. The identified individual and environmental
resources could be used to reduce the amount of stress, and
result in increased family life satisfaction and a higher
quality of life.

In this study, economic well-being which was measured
in objective and subjective terms, was another predictor of
one's evaluation of quality of life. In previous studies,
objective income alone explained only a small amount of
variance in quality of life (Inglehart 8 Rabier, 1986).
However, this study provided a more complete idea of how
families evaluate their level of economic well-being. It

was found that economic well-being was based on not only the

105
objective measure of per capita income, but also on its
adequacy in relation to family needs and goals.
Additionally, changes in the family financial situation over
time played an important role in determining economic well-
being. Educational programs designed to teach family
resource management skills, and efficient use of resources

are recommended.

APPENDICES

APPENDIX A

106

 

 

new; anN m8: ON“ . o: .- n2 .- hnNc thf 96.- “mam
:5; a3. Z on— . a8... NON. hNN. _nN. 3N. NNo. n38
boo." ocean $6 $8 an. 8N. 02. EN. n8: 30w
to. :56 N8. :0: FN. 3n. 3N. 8N. «Nat 320m
2..— A one“ n3. N3 .- 3N. onN. 03.. non. oNor nun—mm
”a. and 96. N— .- 53. can. Nhn. hub. 39.- gm
befig _: .N :6. woo. 89- «8. 23. N3. ONN. mam—P43
8a. uNaN 9 n. 90.- §. 36. 03: N8. 96. “ammo
hub. hood one. she.- voo. «3. a8. 2b. as. nun—<—
3». G" .n SN. £3.- a8.- «8: who.- ONOK 8°. 3E0
New; SEN 89— 08.- 3: . N8. n3. *8. «NO. v30.—
nwv. XEN 084 30.- N8.- one... 3 u .- N3. 3am
Nun; nnN.m 89— gb. NNQ. N8. ac.- 30m
094 8nd 8?— New. (Nb. 586 Slow
«he; 3nd 804 E. . m8: Ema
NnNé main 08." v8: gm
39.n— SNdn g; mam—Pa:
.D.m am: 122 mag» 30% QIUm nun—mm gm

 

numb-=—

esouaSoG peace-Sm cos .382 .scomocmoou cows—ebony u 033—.

 

o8.— cocr owe; 39.- N2 .- 8— . N3: N86 ace: 3?- nus—Fm
7 08; New. 2%. wan. N3. 36. 2.— . nnN. ms. 300
nlu con: can. 8n. 2%. «no. cue: «8. N8: 38m
89— 03.. won. NO. _NO. 6ND. n8: 330m
08; ans. One. who. ”no. —No.- 2mm
89— N3. one. n8. «5: max—mm
Rx: No. n3. «no. mam-FA:
95.— Na. can. .3th
804 N3. mun—<—
§.— mam-.5
mung—Fm n38 mam—Um 310m nus—mm 3mm mam—Pd.— mwmmu mun—<— was

 

 

A9258 5 035.

APPENDIX B

QUESTIONNAIRE (1983)

1138

 

He would like to have some background information about your family to
help us in our study. Please fill in the following information about
each member of your household. identifying each person by their relation-
ship togyou. _

 

—

 

First, think about yourself.
1Sex: h F No. 8 Yr. of Birth_ Yrs. of School Completed
(“"1" we) Marital Status If Married, Ho. 1 Yr.0f Marriage

 

Next, think about each of your children. starting with the oldest child.
He will be asking questions about each of your children later in this
questionnaire. Please make sure your answers are from oldest to youngest
in each instance.

 

 

Birth Yrs. of Living at (If N0) Date I of

 

 

 

 

Date School Home Reason for Left Support
QZ Sex Ho. Yr. Cowleted Yes or he Leaving lb. Yr. You Provide
a. Child 1 H F
b. 22119.? H F
c. Child 3 H F
d. Child 4 H F
e. Child 5 h F

 

f. Child 6 H F
(add on if necessary)

 

IfFinally, think about each other member of your household:_]

 

Yrs. of S of
Relationship Sex Birth Date School Marital Support

3 to You H or F No. Yr. Completed Status You Provide

 

 

 

 

Religious Preference:
04 Catholic Protestant Jewish Other (please specify)

 

 

 

 

109

05 which of the following best describes your racial or ethnic identification?
Black Hhite (Caucasian)
Chicano (Mexican-American) Oriental
Native American (American Indian) Other (please specify)

 

Q6 How many years have you lived in your present comnity?
07a. what is the size of the comnity in which you live? (circle one)

1. Less than 2,500 persons and outside an urbanized area
2. More than 2,500 persons but less than 50,000 persons
3. 50,000 or more persons

6. (If you live in a comnity of less than 2.500 persons) Is your home
on less than 1 acre of land or on a city or suburban lot?

1. Yes (skip to 08)
2. Mo

 

c. (If no) Do you live on a farm?

1. Yes
2. Mo (skip to 08)

a. (If you live on a farm of more than 1 acre) Did your farm produce _
$1,000.00 or more in sales of crops, livestock. or other farm products
during the preceding year?

1. Yes
2. Mo

 

Please read each of the events listed below and mark whether it was ex-
perienced by any family member in the last three ears. If yes. please
circle the number showing how disturbing t was a n icate whether it
occurred in the last twelve months.

Mow Oi sturbing Mas

 

This Event?
Has This Event a >- '
08 rmu use evens Happened to " - " 014 it
Your Family in g 5 g g Occur In
A. Internal to the The Last Three .- - a -- 0’: The Last
Family Years? 2 a: - 3 u 12 Months?
a. Death of a menber Yes No 1 2 3 4 5 Yes Mo
b. Marriage of a member Yes he 1 2 3 4 5 Yes he

111)

How Disturbing Has
This Event?

Has This Event Z: ,.
Happened To ‘ 2'- 3 5? Did It
Your Family In 5 § u u- Occur In
The Last Three -— ... ‘5‘ L'- 95 The Last
Years? 3 8% 2 3 3‘1 12 Months?
. Member moves out of
home (for independence.
for added schooling,
for job. for marriage) Yes No I 2 3 4 5 Yes No
. Member moves back
(unemployed. divorced.
or separated. etc.) Yes No I 2 3 4 5 Yes No
Mon-member (renters.
boarders. etc.) moved .
into home . Yes No I 2 3 4 5 Yes No
Marital separation '
occurs Yes Mo 1 2 3 4_ 5 Yes No
Periodic absence of
family eeuiber due to .
' work demands Yes he 1 2 3 4 5 Yes he
Family pet dies Yes he 1 2 3 4 5 Yes he

Pregnancy of unmarried .
member Yes he 1 2 3 4 5 Yes No

. Member demanding of new

privileges. exemptions

from family rules.

choice of friends.

dates. etc. Yes Mo 1 2 3 4 5 Yes No

Adult child has
trouble achieving

independence Yes M0 1 2 3 4 5 Yes No
. Household chores pile

up Yes No 1 2 3 4 5 Yes No

Family took a

stressful vacation Yes No _ 1 2 3 4 5 Yes No

This Question Continues On The Next Page

JJll

Mow Disturbing was
This Event?

 

Mas This Event > a 5 Did It
happened To .a 0-
Your Family In ’5 5 g g Occur In
The Last Three k- .-. ‘5‘ 5 $5- The Last
Years? 2 d 2 a 3‘; 12 Months?
8. Family, School and Mark
. Member drops out of
school before com-
pleting training Yes No 1* 2 3 4 5 Yes No
. Member returns to
school after time
away Yes Mo I 2 ‘ 3 4 5 Yes No
. Major wage earner
loses or quits Job Yes No I 2 3 4 5 Yes No
. Major wage earner
starts or returns
to work Yes Mo 1 2 3 4 5 Yes Mo
. Menber given promotion Yes Mo 1 2 3 4 5 Yes Mo
. Member changes to new
job or shifts career Yes Mo 1 2 ’ 3 4 5 Yes No
. Major wage earner
retires from work Yes No I 2 3 4 5 Yes No
Member accepts time
consuming, unpaid
assignment in volun-
tary association
(scouting. church. or
service agency) Yes Mo 1 2 3 4 5 Yes Mo
. Outside activities
draw adult members
away from family Yes Mo 1 2 3 4 5 Yes No

. Member's hours!
scheduling of work
change Yes Mo 1 2 3 4 5 Yes Mo

cc.

ff.

Member has major
conflict with boss
and/or others at work

C. Family, Relatives and
Close Friends

. Relatives/in-laws become

intrusive (offer un-
welcome advice. gifts)

Death of husband's or
wife's parents

Death of brother or
sister

. Death of close friend

and confidant

Married children
“freeze out“ parents

Member breaks up with
close friend or
confidant

. Relative dies (not

parent or sibling)
D. Family and Health

Major wage earner
experiences serious
illness or accident

Member experiences
serious emotional
probl ems

1112

Has This Event
Happened To
Your Family In
The Last Three
Years?

Yes No

Yes Mo
Yes No
Yes No
Yes Ho

Yes No

Yes Ho

Yes No

Yes Ho

Yes Mo

How Disturbing Has

This Event?

NOT

1

SLIGHTLY

N

2

MODERATELY

QUITE

3

b

4

EXTREMELY '

U!

5

Did It
Occur In
The Last
12 Months?

Yes No

Yes No
Yes No
Yes No
Yes No_'

Yes No

Yes Ho

Yes NO

Yes No

Yes Mo

This Question Continues On The Next Page

hh.

11.

JJ-

11.

PP-

1153

Has This Event
Happened To
Your Family In

The Last Three
Years?
Child member experi-
ences serious
illness/accident Yes Ho
Aged parent(s) becomes
seriously ill or
disabled requiring
direct care Yes Mo
Member experiences
menopause Yes Mo

. Aged parent committed to

institution or placed
in nursing home Yes Ho

E. Family, Household Finance
and the4Law

Husband's or wife's parents
or siblings require
financial assistance Yes Ho

. Cut in total family

income Yes No

. Expenses exceed total

family income requiring
going into debt Yes Mo

. Family takes a major loss

in stock market, bank
failure, bad debts. etc. Yes No

Family receives windfall

funds (inheritance.

lottery win. or other '
unanticipated gain) Yes Ho

How Disturbing Has
This Event?

MOOERATELY

QUITE
EXTREMELY

SLIGHTLY

MOT

pm
N
U
‘
U'I

Did It

Occur In
The Last
12 Months?

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Ho

Ho

cI‘l-

It.

fit.

an.

Member starts receiving
public assistance in
the form of food stamps.
rent subsidy or AFDC

Member takes out or
refinances a loan to
cover increased
expenses

Family meter involved
with courts; robbed or
assaulted. arrested for
crime or minor mis-
demeanor. jailed. or
involved in lawsuit

Family forced to dip
heavily into family
savings

Member taking on
additional jobs

Member experiencing
demotion. job bumping.
or retooling

 

 

2114

Has This Event
Happened To
Your Family In
The Last Three

Years?

Yes

Yes

Yes

Yes

Yes

Yes

F. Other Events Hot Covered

Yes

Yes

No

Ho

Ho

No

No

How Disturbing Has
This Event?

HODERATELY

SLIGHTLY
QUITE

NOT
EXTREMELY

H
N
u
&
U!

 

[*Please Go On To The Next Page]

 

Did It

Occur In
The Last
12 Months?

Yes

Yes

Yes

Yes

Yes

YES

Yes

Yes

Ho

Ho

Ho

HO

Ho

NO

115

 

He're interested in the health of each member of your family. Please use
the codes given below to indicate how often the following items apply to

menbers of your family.

1 Never

2 Seldom

3 Sometimes

4 Frequently

5 Almost Always

For example. if child I smokes “frequently“ and child 4 smokes
”sometimes" and no one else in the family smokes. then you would answer:

Oldest ——b Youngest

“*“w m 13m

 

 

 

on How often have medaers of
your family:

a. had trouble sleeping...........: ....... '.
I». had accidents............-. ..........
c. been irritable
«I. been depressed
a. smoked cigarettes. cigars. or a pipe....
I. used prescription drugs .................
3. had a weight problem ....................
b. used alcohol ............................
1. found it difficult to relax .............
1. had headaches ............. . .............

1:. had muscle tension. nervous
indigestion or anxiety ................

1. had colds or flu ........................

 

JLLG

025 In general. which of the following best describes any changes in your total
family income over the past 3 years

a. increased more than 25%

b. increased 5 to 251

c. changed less than 51 (plus or minus)

d. decreased 5 to 251

e. decreased more than 25!

f. fluctuated up and down over the 3 years. .

026 To what extent do you think your income today is enough for you to live on?

a. can't buy some necessities

b. can meet necessities only

c. can afford some of the things we want but not all we want
d. can afford about everything we want

e. can afford about everything we want and have some left over

027 Thinking about your family's overall financial condition -- what you own.
owe. earn. are able to buy. and so on -- which of the following best
describes any change in your overall financial condition over the 3st

3 Ears?

a. mach worse

b. worse

c. same (skip to 029)
d. better

e. much better

023 If your financial condition has changed during the 29st 3 years. please
describe the change(s) below.

 

 

 

For each item listed below. think about the amount of money your family
spends for family members now living in your household. Over the st
3 ears. how has the amount you spend changed? In general. 33 you Yeel
that tile amount you spend today is:

Lot Mo Lot
029 Less Less Change More More
a. Food eaten at home 1 2 3 4 5
b. Food eaten away
from home 1 2 3 4 5
c. Clothing purchases 1 2 3 4 5
4. Clothing repairs
and alterations 1 2 3 4 5

f. Hopeful l 2 3 4 5 6 7 Oiscouraging

g. Rewarding 1 2 3 4 5 6 7 Disappointing

h. Brings out the Doesn't give me
best in me 1 2 3 4 5 6 7 much chance

 

Please circle the number which best describes how satisfied you are with
your life as a whole.

 

 

 

Q42 Completely Completely
Dissatisfied I 2' 3 4 5 6 7 Satisfied

Please circle the number which best describes how satisfied you are with
your family.

Completely Completely
043 . Dissatisfied Satisfied

a. Your family
. life 1 2 3 4 5 6 7

_ b. Your relation-
ship with your
spouse 1 2 3 4 5 6 7

c. Your relation-
ship with your

children 1 2 3 4 5 6 7
a. The relation-

ship your

children have

with each other I 2 3 4 5 6 7

 

your marriage. The middle point represents “happy“.

 

Please circle the number which best describes how happy you are with ]

 

044 Extremely I Extremely
Unhappy 1 2 3 4 5 6 7 Happy

APPENDIX C

QUESTIONNAIRE (1985)

2118

 

He would like to have some background information about your family to help us in our
study. He are especially interested in what has happened in the last two years. Please
circle appropriate letter or fill in the blank with your answers. First think about
yourself.

(21

()2
(23
(36

05

06

07

QB

09
QlOa

QIOh

What is your sex?

a. Hale
b. Female

what is the date of your birth? Mo. Yr.

How many years of school have you completed?
What is your religious preference? (Circle the letter)

a. Catholic

b. Protestant

c. Jewish

d. Other (please specify)

4.

Have you moved to a different home in the last two years?

a. Mo
b. Yes

(IF YES please circle the appropriate letter
a. Moved within the same community
b. Moved to a different community

Do you live on a farm?
a. No
b. Yea. less than one acre
c. Yes. more than one acre
(If more than one acre) Did your farm produce 51.000.00 or more
in sales of crops. livestock. or other farm products during the

preceeding year?

a. No
b. Yes

What is your marital status?

 

Has your marital status changed in the last two years? (Please circle)

a. No
b. Yes

(IF YES How has it changed? (Circle more than one category if appropriate)

a. Became widowed c. Became separated or living apart
b. Became divorced d. Remarried Mo. Yr.

1119

 

-Hext. think about each of your children. starting with the oldest child. He will be
asking questions about each of your children later in this questionnaire. Please make
sure your answers are from oldest to youngest in each instance.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

011 Sex Birth Yrs. of Living at Z of support
(Circle) Data School Home You
Male/Female Mo. Yr. Completed Yes or No Provide
a. Child l M E Y M
b. Child 2 M F Y M
c. Child 3 M Y Y H
d. Child 4 M F Y H
a. Child 5 M F Y M
f. Child 6 M F Y H
3. Child 7 M F Y M
h. Child 8 M F Y M
1. Child 9 M Y Y M
1. Child 10 M Y Y M
k. cam 11 x r r a
1. Child 12 M I Y M
L__ He are interested in the changes that have taken place in your household. ‘]

 

012 Have any of your children left home in the last 2 years? (Please circle the letter)

 

 

 

 

 

 

 

 

 

a. No
b. Yes
013 [__.' (IF YES) Please indicate on the chart below. the number of which child
left. the reason for leaving. and the date.
014 Have any of your children returned home to
live with you in last 2 years?
a. No
b. Yes
015 l-.. IF YES) Please indicate on the chart below.
the number of which child returned. the reason.
and the date. v
Child Number Reason for Returning Date Reason for Leaving Date

(as listed above) (Approx) (Approx)

 

 

 

 

 

 

12(7

 

Please read each of the events listed below and mark whether it was experienced by
If yes. please circle the number showing

any family member in the last twogyears.

how disturbinggit was.

 

 

 

 

 

Ql9 EVERY Has This Event Happened How Disturbing Has
to Your Family in.... This Event?
[East 12 mos (Yr. Before s, A
the last 12 mos. 3r -\ ,7
-y u
4" of .4" .3:
9 iv 'b 57
I. IIHEDIAIE EMILY Q9 a? *9 Q I”:
a. Death of a member Yes Ho Yes Mo 1 2 3 4 5
b. Marriage of a member Yes No Yes Mo 1 2 3 4 S
c. Member moves out of
home (for independence.
for added schooling. .
for job. for marriage) Yes Mo Yes Mo l 2 3 4 S
d. Member moves back
(unemployed. divorced.
or separated. etc.) Yes No Yes Ho 1 2 3 4 S
e. Mon-member (renters.
boarders. etc.) moved
into home Yes Mo Yes No l 2 3 6 5
f. Marital separation
occurs Yes In Yes Mo 1 2 3 4 5
g. Periodic absence of
family member due to
work demands - Yes No Yes No l 2 J b 5
h. Family pet dies Yes Mo Yes No l 2 3 A S
1. Pregnancy of unmarried
member Yes No Yes No l 2 3 6 S
J. Member demanding of new
privileges. exemptions
from family rules.
choice of friends.
dates. etc. Yes No Yes No l 2 3 a 5
k. Adult child has
trouble achieving
independence Yes No Yes No l 2 3 4 S
1. Household chores
pile up Yes Mo Yes No I 7 3 A S
m. Family t ok a
atresafuI vacation Yes No Yes No I 2 3 a S

 

’ 121

. EVENT Has This event Happened How Disturbing
to your Family In... "as This Event?

 

 

Last l2 mo Yr. Before the
last 12 mos.

 

II. SCHOOL AND WORK 0

a. Member drops out
of school before com-
pleting training Yes No Yes No l 2

U
0‘
U

a. Member returns to
school after time
away Yes Mo Yes Mo 1 2 3 4 5

p. Major wage earner -
loses or quite job Yes Ho Yes Mo 1 2 3 4 S

 

q. Major wage earner
starts or returns
to work Yes Me Yes Mo 4 I ? 3 4 S

r. Member given promotion Yes Mo Yes Mo I 2 3 4 S

s Member changes to new
job or shifts career Yes No Yes Mo I 2 3 4 S

t. Major wage earner
retires from work Yes Mo Yes Me I 2 3 4

b1

 

u. Member accepts time
consuming. unpaid
assignment in volun-
tary association
(scouting. church. or
service agency) Yes No Yes Mo I 2 3 4 5

v. Outside activities
draw adult members
away from family Yes No Yes Ho I 2 3 4 S

w. Member's hours/
scheduling of work
change Yes Ho Yes Mo I Z 3 4 5

x. Member has major
conflict with boss
and/or others at work Yes No Yes No I Z 3 4 S

 

__"m. afflm's‘fmu cross raifi'bs

y. Relatives/In-Iawa become
intrusive (offer un-
welcome advice. gifts) Yes Mo Yes Me I Z 3 4 S

a. Death of husband's or
wife a parents Yes Me Yes, Mo l 2 3 4 S

 

 

122

 

 

 

 

 

EVENT Has This Event Happened How Disturbing Has
to Your Family.ln... This Event?
[fiEast 12 mos.]Yr. Before the
last 12 mos.
N
3's?
6’34“
«u :7 lb ‘9
set” a
as. Death of brother
or sister Yes No Yes No l 2 3 5
bb. Death of close friend
and confident Yes No Yes No I 2 3 5
cc. Married children "freeze
out" parents Yes No Yes No I 2 3 5
dd. Member breaks up with
close friend or confi-
dent Yes No Yes No I 2 3 5
ea. Relative dies (not parent
or sibling) Yes No Yes No I 2 3 5
IV. HEALTH
ff. Major wage earner experi-
ences serious illness or
accident Yes No Yes No I 2 3 S
gg. Member experiences serious .
emotional problems Yes No Yes No l 2 3 5
hh. Child member experiences
serious illness/accident Yes No Yes No I 2 3 5
ii. Aged parent(s) becomes
seriously ill or disabled
requiring direct care. Yes No Yes No I 2 3 5
5]. Member experiences
menopause Yes No Yes No I 2 3 S
kk. Aged parent committed to
institution or placed in
nursing home Yes No Yes No I 2 3 S
V. HOUSEHOLD FINANCE AND LEGAL ISSUES
ll. Husband's or wife's parents
or siblings require finan-
cial assistance Yes No Yes No I 2 3 5
mm. Cut in total family income Yes No Yes No l 2 3 5
an. Expenses exceed total
family income requiring
going into debt Yes No Yes No I 2 3 5
oo. Famil takes a msiot
loss In stock market.
bank failure. bad debts.
etc. Yes Mo Yes No I 2 3 S

 

123

 

 

 

QZO

OZI

 

EVENT
Has This Even Happened How Disturbing
To Your Family In.... Has This Event?
I Last 12 mos. Yr. Before the] I I
last 12 mos. g A
‘V ‘U a.
.. e
e «M 0 A?
Q 3' 4’ . I.
pp. Family receives windfall e 3 0 s1" 9
funds (inheritance. a. .27 i d 4'
lottery win. or other * "
unanticipated gain) Yes No Yes No I 2 3 4 5
qq. Member starts receiving
public assistance in the
form of food stamps. rent
subsidy or AFDC Yes No Yes No I 2 3 4 5
rr. Member takes out or
refinances a loan to cover
increased expenses Yes No Yes No I 2 3 4 5
as. Family member involved
with courts; robbed or
assaulted. arrested for
crime or minor mis-
demeanor. jailed. or
involved in lawsuit Yes No Yes No I 2 3 4 5
tt. Family forced to dip
- heavily into family savings Yes No Yes No I 2 3 4 5
uu. Member taking on additional
jobs Yes No Yes No I 2 3 4 5
vv. Member experiencing
demotion. job bumping. or
retooling Yes No Yes No I 2 3 4 5
VI. OTHER EVENTS NOT COVERED
ww. Yes No Yes lo I 2 3 4 5
xx. Yes Mo Yes No I 2 3 4 S

 

Of the events just considered that happened in your family during the
last 2 years. which one had the most impact on your family?

 

Hhat resources were important to you in coping with this event?

 

 

.124

 

He're interested in the health of you and the children living at home now. Please use
the codes given below to indicate how often the following items apply to members of your
family.

 

1. Never

2. Seldom

3. Sometimes

4. Frequently

5. Almost Always

For example. if child 1 had trouble sleeping ”frequently” and child 4 "sometimes"
and no one else in the family had trouble sleeping. then you would answer:

s~fiwbbxb
‘b b. ‘b 5, ‘b
‘wb\>~~b ~~N
'» ‘v » b' ., h' a. ‘~
~°$6<P§o°c°¢c°

 

had trouble sIeeping-- l 2 l I 4 I

 

 

 

 

 

 

 

 

 

 

 

 

 

 

027 How often has each family member
living in your household:

 

 

 

 

.9 as as" a?» ”‘5 (33" §¢b§¢b (:95
a. had trouble sleeping .. . .. ..
b. had accidents . .. . .. -- :- .
c. been irritable -- -- - -- .. .
d. been depressed. . .. . .. . .. .

 

e. smoked cigarettes. cigars. or
ar‘peasses-sescaasoaa

 

i. used prescription drugs . .. ..

 

g. had a weight proble. . . .. . ..

 

h. used alcohol- -- - -- - -- -- .

 

i. found it difficult to relax . -

 

1. had headaches .. .. . ., .. ..

 

k. had muscle tension. nervous
indigestion or anxiety

 

I. had colds or flu . .. . .. .. .

 

 

 

 

 

 

 

 

 

 

 

032

033

03:

035

036

037

030

‘125

Do you have any other fob in addition to your main job?

 

.a "O
b. Yes
(IF YES) "hat is your job title?
Hhat is the attitude of your spouse toward your additional job? (Please
circle the number using the following scale).
NOT .
- I 2 3 4 5 6 7 VERY
SUPPORTIVE SUPPORTIVE

Are you currently in the military Service?

a. No

b. Yes ---1

(IF YES ) Please circle the appropriate answer.

a. Employed full-time member of the Army.

b. Employed full-time member of a military service other than the army.
c. Part-time member of Army Reserve or Army National Guard Unit.

d. Part-time member of a military service other than the Army.

Next. think about your total family income for 1984 received by you and all family
members who live with you. This is the total income before taxes. Be sure to
include all sources of income such as earned income. investments. social security.
your own business. job-related benefits. welfare benefits. and so on. If your family
forms or has its own business. indicate net farm or 523 business income before taxes.
(Clrcle'tha letter of the category that is closest to your total family income.

a. $4,999 or leg! k. 550,000-354,999
h. $5.000-59,999 I. 355,000-559.999
c. 310,000-314,999 I. 360,000-364,999
d. 315,000-519,999 n. 365,000-S69.999
e. 320,000-524,999 o. 570,000-374,999
f. 525,000-529,999 p. 575,000-579,999
g. 530,000-534,999 q. 580,000-384,999
h. 335,000-539,999 Y. 585,000-589,999
l. $40,000-S44,999 s. 590,000-595,999
j. 345.000-549.999 t. 595.000 and up

Now. think about your personal income before taxes for 190?. Be sure to include all
sources of income that you receive personally: such as earned income. investments.
social security. your own business. Job-related benefits, welfare benefits. rent and
so on. If you farm or have your own business. we would like to indicate your net

farm or net business income before taxes.

s. No personal income I. 350,000-354,999
b. $4.999-or less m. 355,000-359,999
c. $5.000-39.999 n. 360,000-564,999
d. 310,000-514,999 o. 365,000-369,999
e. 315,000-319,999 p. 370,000-574,999
f. 520,000-374,999 q. 375,000-579,999
3. 325,000-329,999 r. 380,000-384,999
hf 330,000-534,999 I. 505,000-309,999
0 s ' s E. s s " s
3. 323.333-222.33: .. gamma:
k. 545,000-349,999 v. $100,000 and up

1265

Q39 If you had a family business or farm. did you experience a net loss in the last year?
(Please circle the letter).

a. No
b. Yes .
c. Question does 22E apply to me.

Q40 In general. which of the following best describes any changes in your total family
income over the past 2 years?

a. increased more than 252

b. increased 5 to 25!

c. changed less than 51 (plus or minus)

d. decreased 5 to 252

a. decreased more than 252

f. fluctuated up 221 down over the 3 years.

Q41 To what extent do you think your income today is enough for you to live on?
a. can't buy some necessities
b. can meet necessities only
c. can afford some of the things we want. but not all we want
d. can afford about everything we want
a. can afford about everything we want and have some left over

Q42 Thinking about your family's overall financial condition -- what you own. owe. earn.
are able to buy. and so on -- which of the following Egg; describes any change in
your overall financial condition over the past 2 years?

a. much worse

b. worse

c. same (go to.question 44)
d. better '

e. much better

Q43 If your financial condition has changed during the past Zgyears. Please describe
the change(s) below.

 

Q44. Did a family member take a second or third job?

a. No
b. Yes

045 Did a family member work overtime?

a. No
b. Yes

Q46 Did a family member. who was not working. go to work?

a. No
b. Yea

 

 

127

 

Here are some words and phrases which we would like you to use to describe hoggyou feel
about your present life. Please circle the number which best describes where you stand

in describing your feelings.

 

 

 

Q57 a. Boring I 2 3 4 S 6 7 Interesting
° b. Miserable 1 2 3 4 s 6 7 Enjoyable
c. Useless I 2 3 4 5 6 7 Horthwhile
—T.._—Ilnfriand1y 1 2 3 4 s 6 7 'FLriendly
e. Empty I 2 3 4 5 6 7 Full
f. Diacouraging I 2 3 4 5 6 7 Hopeful
6

g. Disappointing I 2 3 4 5 7 Rewarding

h. Doesn't give
me much chance I 2 3 4 5 6 7 Brings out the best in me

 

Please circle the number which best describes how satisfied you are with each of the

following:

050 Completely Completely
Dissatisfied Satisfied
a. Your life as a whole I 2 3 4 5 6 7
b. Your family life . I 2 3 4 S 6 7

c. Your relationship with your
spouse I 2 3 4 S 6 7

d. Your relationship with your
children I 2 3 4 S 6 7

e. The relationship your
children have with each other I 2 3 4 5 6 7
Please circle the number which best describes how happy you are with your marriage.

The middle point represents "happy”.

059 Extremely Extremely
Unhappy I ~2 3 4 5 6 7 Happy

 

BIBLIOGRAPHY

BIBLIOGRAPHY

Abbey, A.. & Andrew. F. (1986). Modeling the psychological
determinants of life quality. In F. M. Andrews (Ed. ),

Research on the 93; lity g: lite. Ann Arbor: Survey

Research Center, Institute for Social Research, The
University of Michigan.

Ackerman. N.. & Paolucci. B. (1983). Objective and
subjective income adequacy: Their relationship to
perceived life quality measures.

B§§§§IQDI_12.25“43-

Alston. J.P., Lowe. 6.0.. a Wrigley. A. (1974). Socio-
economic correlates for four dimensions of self-

perceived satisfaction. flungn_gzganiza§ign, 33. 99-102.

Anderson. J. C.. & Gerbing. D. W. (1988). Structural
equation modeling in practice: A Review and recommended

two-step approach. Psychological_aulletin. 103. 3.
411-423.

Andrews, F. M. (1981). Subjective social indicators.
objective social indicators, and social accounting
systems. In F. T. Juster & K. C. Land (Eds. ). figgial

agggunting_§y_tgm§, 377-419. New York: Academic Press.

Andrews. F. M.. & Bubolz. M.. & Paolucci. B. (1980). An
ecological approach to study of the family. Marriage
and_£anilx_3exiew§. 3 Spring lsummerillzl

Andrews. F- M.. & Wither. 8- (1976). Social_indicator§_ofi
gg11;bg1ng. New York: Plenum Press.

Antonovsky. A- (1980). Health_atreaa_and_coning. Washington.
D.C., Inc.

Bagozzi. R. P., & Phillips, L. W. (1982). Representing and

testing organizational theories: A holistic construal.
. 459-489.

128

BI

Bu

Bu

129

Bagozzi, R. P., & Yi, Y. (1988L On the evaluation of

structural equation models. 1ournal_9f_the_agadem¥_of
Marketing_fisienoel_1§ 74-94-

Bentler, P. M., & Bonett, D. G. (1980). Significance tests
and goodness of fit in the analysis of covariance

structures. 2sxgbologisal_§ulletin1_§§ 588- 606-

Berry, R., & Williams, F. (1987). Assessing the
relationship between quality of life and marital and
income satisfaction: A path analytic approach. Journal

of_uarriagg_and_the_£amilxl_12. Feb. 107-116-

Bohrnstedt, G. W., 8 Carter, N. T. (1971). Robustness of in
regression analysis. In H. L. Costner (Ed).

Sosiologisal_nethodolosx 118- -146- SanFranciseo=

Jossey-Bass.

Bollen. K- B. (1989)- Stru2tural_egnations_uith_laten§
yariaplgg. John Wiley & sons, Inc.

Bradburn, N. H., & Caplovitz, D. (1965). Beng;tg_gn
happiness. Chicago: Aldine.

Brandtstadter, J. (1984). Personal and social control over
development: Some implications of an action perspective
in life-span developmental psychology. In P. B. Baltes,
& O. G. Brim, Jr. (Eds. ),
bgnayigz (Vol. 6, 1-32L New York: Academic Press.

Braun, P. M. W. (1977). Psychological well-being and
location in the social structure (Doctoral
dissertation, Texas Woman's University, 1979).

D1ssertat1on_Abstraots_lnternational. 38. 2351A.

Brown, G. W., & Andrews, B. (1986). Social support and
depression. In M. H. Appley & R. Trumbull (Eds. ),
I !.II _: .,.A.. o: 1.0.71.1!

sogial_nersn_stixes (pp. 257-282) New York: Plenum.

Browne, M. W. (1982). Covariance structures. In D. H.
Hawkins (Ed. ), e
Cambridge: Cambridge University Press, pp. 72- 141.

Bubolz, N. M., Eicher, J. B., Evers, S. J., 8 Sontag, S.
(1980). A human ecological approach to quality of life:
Conceptual framework and results of a preliminary

study- S22111.1nd12ators_3e§ear§h. 7- 103-136-

Bubolz, N. M., Eicher, J. B., & Sontag, S. H. (1979)._ The
human ecosystem: A model. 12urnal_2f_nome_fisonomigs1
21 (1), 1123-31.

130

Bubolz, M. M., & Sontag, M. S. (In pressL Human ecology
theory. In P. Boss, W. Doherty, R. LaRossa, W. Schumm,

& s. steinmetz (Eds ). Eour2ebook.of_familx_tneorie§
and_meth2Qs1__A_Qontextual_annzoagh. New York: Plenum

Press.

Bubolz, M. M., Walker, R., 8 Lee, M. P. (1989). Family and
economic stressors and quality of life: A rural-urban
comparison. fissislggissl_Abs§1ss;s. North Central
sociological Association.

‘ Campbell, A., Converse, P., & Rodgers W. (1976). Ins_ggsligy
EQSIEIQQLIQDE- New York: Russell Sage Foundation.

Campbell. A- (1981). Ine_sense_of_well:being_1n_bmezisa.
New York: McGraw-Hill.

Campbell, A. & Converse, P. E. (1980). Ths_guslity_sf
Ams;issn_lifss_121s. Ann Arbor: The University of
Michigan, Inter-University Consortium for Political and
Social Research.

Campbell, D. T., 8 Fiske, J. D. (1959). Convergent and
discriminant validation by the multitrait-multimethod

matrix. Rsxsnologioal_nnlletin1_§§ 81-105-

Cantril. 8- (1965). The_nattern_of_hnnan_gon2erns- New
Brunswick, NJ.: Rutgers University Press.

“Caplan, R. D. (1983). Person environment fit : Past ,
present, and future. In C. Cooper (Ed.), fitgsss
° ' (pp. 35-78).

London: Wiley.

Caplan, R.D., Cobb, 8., French, J. R. P., Jr., Harrison, R.
V., & Pinneau, S. R., Jr. (1980). ng_dsmsngs_snd
w9rker_health1_na1n_effests.and_oesunat12nal

. Ann Arbor: Institute for Social Research,
The University of Michigan.

Clemente, F., 8 Sauer, W. J. (1976). Life satisfaction in
the United States. fissis1_£szssss_§1, 621-631.

Cohler, B. J., Boxer, A. M. (1984). Middle adulthood:
Settling into the world-person, time, and context. In

0- Offer 8 H- Sabshin (Eds-I. Normalitx_and_life_sxgle
(pp. 145-203). New York: Basic Books.

Cumming, E-.& Henry. N- E- (1961). Growing_oldi_1he_nrosess
of_disengagement

. New York: Basic Books.

131

Davis, E. P., 8 Helmick, S. (1985).Family financial
satisfaction: The impact of reference points. gems

Es2nom12s_Besearsh_lournall_11. 123-131-

Deacon, R., 8 Firebaugh, F. (1988). Esm11y_;sssgzss
msnsgsmsng. Needham Heights, MA: Allyn 8 Bacon.

Dohrenwend, B. S., 8 Dohrenwend, B. P. (1974). strsssfiul

l1fs_exsnts1_Their_nature_ans_effsgts- New York. John
Wiley.

Edwards, N. J., 8 Klemmack, D. L. (1973).Correlates of life
satisfaction: A re-examination. lggznsl_sfi

Gerontolosxl_za. 497-502-

Estes, R. J., 8 Wilensky, H. L.(1978). Life cycle squeeze
and the morale curve. fissisl_£zsp1sms, 25, 277-292.

Fiske. N- (1979)- u1ddle_ase1_1he_nrime_of_lifez New York:
Harper & Row.

Fletcher, C., 8 Lorenz, F. (1985). Structural influences on
the relationship between objective and subjective
indicators of economic well-being. figsisl_Ingiss§szs
Bsssszsh, 16, 333-345.

Freudiger,P. T.(1980). Life satisfaction among American
women (Doctoral dissertation, North Texas State
University. 1979). _Dissertat19n_Abstrasts
Ingsznsgisnsl, 40, 6438A. (University Microfilms No.
80-12,882)

Gastil, R. (1970) Social indicator and quality of life.
2nhl19_Administration_3exiew1_19. 596-601-

Glenn, N. D. (1975). The contribution of marriage to the
psychological well-being of males and females. isnnnsl
9f_narria9e_ann_the_£amilxi_1l. 594-600-

Glenn, N. D., 8 McLanahan, S. (1981). The effects of
offspring on the psychological well-being of older
adults. J9urnal_of_narriage_an§_tne_£amilx1_11. 138-
150.

mGlenn, N. D., 8 Weaver, C. N. (1981). The contribution of
marital happiness to global happiness. isgggsl_sfi

aarriase_and_ths_£amilxl_11. 161-168-

b Glenn, N. D., 8 Weaver, C. N. (1979). A note on family

situation and global happiness. §Q§1Q1_EQI£§§121. 960-
967.

132

Goodstein, J.,Zautra, A., 8 Goodhart, D. (1982). A test of
the utility of social indicators for behavioral health

service planning. S9cial_1ndicators_3esearch. 10. 273-
295.

Griffore, R. J., 8 Phenice, L. A. (1988). Causality and the

ecology of human development- 112.2522n21291251_3g22:g1
15, 515-525.

Gurin, G., Veroff, J., 8 Field, S. (1960). Amsxissgs_yisg
thsi;_mss§s1;nsslgn. New York: Basic Books.

Guttmann, D. (1978). Life events and decision making by
older adults. Ine_§erontolosistl_1§. 462-467-

Haavio-Mannila, E. (1971). Satisfaction with family, work,
leisure and life among men and women.

21.

Hafstrom, J., 8 Dunsing, M. (1973). Level of living:
Factors influencing the homemaker’s satisfaction. figms

Ec2n2m12s_Besearcn_lournall_zl_21_nes-

Harry, J. (1976). Evolving sources of happiness for men over
the life cycle: A structural analysis. anrnsl_gfi

Abn2rmal_and_§221al_£sycholosy. 29. 289-296-

Havighurst, R., 8 Albrecht, R. (1953). glgsr_psgp1s.
Logmans, Green, New York.

Herzog, A. R., 8 Rodgers, W. L. (1986). Satisfaction among
older adults. In F. M. Andrews (Ed. ), Bssssrsh_sn_§hs_
92.11ty_of_life (pp. 235-251). Ann Arbor: Survey

Research Center, Institute for Social Research, The
University of Michigan.

Hill, R. (1958). Generic features of families under stress.

529151.95sewozk1_12. 139-150-

Holmes, T. H., 8 Rache,R. H. (1967). The social readjustment
scale. I9urnal_of_2syenosomat12_8esearsh1_11. 213-218-

Howell, R. D. (1987). Covariance structure modeling and
measurement issues: A note on interactions among a

channel entity' s power sources. lsg:ns1_s§_flszks§ing
Research1_21. 119- 126.

Hunter, S., 8 Sundel, M. (1989). An examination of key
issues concerning midlife. In S. Hunter, 8 M. Sundel
(Eds.), °
implissgigns. Sage Publication.

133

Inglehart, R., 8 Rabier, J. R. (1986). Aspiration adapt to
situations-But why are Belgians so much happier than
the French? In F. M. Andrews (Ed. ), Essesgch on tbs

gusli_x_gf_lifs (pp. 1- 56), Ann Arbor: Survey Research
Center, Institute for Social Research, The University

of Michigan.

Joreskog, K. G. (1973). A general method for estimating a
linear structural equation system. In A. S. Goldberger
and 0. D. Duncan (Eds. ),

51ructural_Eguation_nerls_1n
gng_sgg1g1_s91gnggs.New York: Academic Press, 85-112.

Joreskog, K. G., 8 Goldberger, A. S. (1972). Factor analysis
by generalized least squares. Esysnsmsggikss_;1. 243-

260.

Joreskog. K. G-. 8 Sorbom. D- (1988L LISBEL_2111_A_99192_22
tne_program_ang_apnlications Chicago: SPSS. Inc-

Joreskog, K. G., 8 Sorbom, D. (1989). L1§BEL_1;_UssrLs_
Isfisxsnss_ggi_s. Scientific Software, Inc.

Mooresville, IN.

Kamman. R. (1982)- 2ers2nal_c1r2nmetances_and_life_eyents_a§
n2or_nregistors_of_nanniness

. Paper presented at the
90th Annual Convention of the American Psychological
Association, Washington, DC.

Kahn, R. L., Wolfe, D. M., Quinn, R. P., Snoek, J. D., 8
Rosenthal, R. A. (1964).

1n_role.conflict.and_ambisnity. New York: Wiley-

Kanner, A. D., Coyne, J. C., Schaefer, C., 8 Lazarus, R. S.
(1981). Comparison of two modes of stress measurement:
Daily hassles and uplifts versus major life events.

, 1-39.

Kmenta. J- (1971). Elements_9f_economstricsl_usw_xorki
uacm1llan_znblisnins_comnan21

Kratzer, C. Y. (1991). Perceived economic well-being or
three cohorts of rural female household financial
managers. (Doctoral dissertation, Michigan State
University, 1991).

Larson, R. (1978). Thirty years of research on the
subjective well-being of older Americans. Jgnnnsl_sfi

EEIQDIQIQ§¥1_12. 109-125-

LBV1DSOD. D. (1973)- Ihe_seasons_of_nanis_life--Kn0pf. New
York.

134

Liang, J., Kahana, E., 8 Doherty, E. (1980). Financial
well-being among the aged: A further elaboration.

l9urnal_of_§erontolosy1_2§. 409-420-

Mammen, S., Helmick, S. A., 8 Metzen, E. J. (1983).
Financial well-being and life satisfaction. In M. M.

Dunsing (Ed.), Ezgceedings sf she sympgsigm gn
nerce1y_d_econ2m1c_well_be1ns

e - ' . Urbana, IL: Agricultural
Experiment Station, 8-13.

Mancini, J.A., 8 Orthner, D.K. (1980). Situational
influences on leisure satisfaction and morale in old

age- l2urnal_of_the_American_§eriatrics_§ociety. 28.
466-471.

Markides, K. S. 8 Martin, H. W. (1979). A causal model of
life satisfaction among the elderly. gsgzns1_s£

QQIQn§9129¥1_11. 36-93-

Mason, R. G., 8 Faulkenberry, G. D. (1978). Aspirations,
achievements, and life satisfactions. fissisl_1ngisstsrs

Researcn1_§. 133-150-

McCubbin, H., Patterson, J. M., 8 Wilson, L. R. (1987).
FILE family inventory of life events and changes. In
H. McCubbin 8 A. I. Thompson, (Eds.),

' . Madison: The

University of Wisconsin.

McMahon, B. J. (1988). Mid-life married women's perception
of the retirement life stage: An exploratory study with
implications for pre-retirement education and home
economics/human ecology. (Doctoral dissertation,
Michigan State University, 1988).

Medley, M. L. (1980). Life satisfaction across four stages

of adult life. 1nfernat1onal_Iournal_of_bsins_and_nnman
Dsyelonment. 11. 193-209.

Michalos, A. C. (1983). Satisfaction and happiness in a
rural northern resource community.

BEEEQIEDL_11. 225-252-

Michalos, A. C. (1982). The satisfaction and happiness of
some senior citizens in rural Ontario. fissisl

Indicator§_3esearch1_1l. 1-30.

Michalos, A. C. (1980). Satisfaction and happiness, fissisl
Indicators_Besearch1_§. 385-422.

135

Miller, M. L. (1980). Adaptation and life satisfaction of
the elderly (Doctoral dissertation, Boston College,

1980L D1ss_rtat12n_Abstreet_lnt_znationall_1l.7367B-
(University Microfilms No. 80- 16 ,611)

Mitchell, R. M. (1976). Paths to happiness: Residence
locality and interpersonal relationships (Doctoral
dissertation, University of Notre Dame, 1976).
D1ssertat1on_Ahstracts_lntsrnational1_11. 73678-
(University Microfilms No. 80-16,611)

Moen, P., Kain, E., 8 Elder, G. (1983). Economic conditions
and family life: Contemporary and historical
perspectives. In R. Nelson, 8 R. Skidmore (Eds.),

' s . 213-259. National
Academy Press. Washington, D.C.

Monat, A. 8 Lazarus, R. (1977). sggsss_sng_ssping. New York:
Columbia University Press.

Morgan. D- (1985)- The_fam1ly1_n9litics_and_social_thegry-
London: Rontledge, Kegan Paul.

Morgan, J. N., 8 Smith, J. D. (1969). Measures of economic
well-offness and their correlates. Amszissn_£ssnsmis

Bsyiew1_§2. 912-926-

Near, J. P., Rice, R. W., 8 Hunt, R. G. (1978). Work and
extra-work correlates of life and job satisfaction.

Academy_9f_nanagement_louznall_21. 248-254-
Nunnally, J. C. (1978). Bfi!§h9fl£§li§.§hé9£¥. 2nd ed. New

York: Russell Sage Foundation.

Olsen, J. K. (1980). The effect of change in activity in
voluntary associations on life satisfaction among
people 60 and over who have been active through time
(Doctoral dissertation, University of Maryland, 1979).
D15sertat1on_Abstract_International. 40. 5211A-
(University Microfilms No. 80-07, 107)

Palmore, E. (1979). Predictors of successful aging. Ins
gerontologistl_12. 427-431.

Palmore, E., 8 Luikart, C. (1972). Health and social
factors related to life satisfaction. lggznsl_sfi

Heal1n_and_fiocial_nshayiorl_11. 68-80-

Perlin, L. I., Lieberman, M. A., Menaghan, E. G., 8 Mullan,

J.T. (1981). The stress process. gsgzns1_gfi_fiss1tn
and_§ocial_nehayiorlzz. 337-356-

136

Rabkin, J. G., 8 Streunig, E. I. (1976). Life events,
stress, and illness. 821en221_121- 1013-1020.

Reich, J. W., 8 Zautra, A. (1981). Life events and personal
causation: Some relationships with satisfaction and

distress. 12nrnal_9f_2ers2nal1ty_and_§ocial_2sysnolog21
51, 1002- 1- 12.

Reich, J. W., 8 Zautra, A (1983). Demands and desires in
daily life: Some influences on well-being. smszissn

12urnal_2f_community_2sycholosyl_11. 41-58-

Riddick, C.C. (1980). The life satisfaction of retired and
employed older women: A re-examination of the
disengagement theory (Doctoral dissertation,
Pennsylvania State University, 1980).

abstracts_lnternational. 41. 2327A- (University
Microfilms No. 80-24, 483). -

Roth, T., Kramer, M, 8 Roehrs, T. (1976). Mood before and
after sleep. '

QI_Q§§Q!§1_1. 123-127-

Schaie, K. W., 8 Willis, S. L. (1986). aggls_gsyslspmsnt
sns_sg1ng (2nd ed.). Boston:Little, Brown.

Schumn, W., McCollum, E., Bugaighis, M., Jurich, A.,8
Bollman, S. (1986). Characteristics of the Kansas
family life satisfaction scale in regional sample.

Efi¥£h21991§31_3£292§§1§§o 975-980-

Selye, H. (1956). Ihs_s;:sss_sfi_11fis. New York: McGraw-Hill.
Selye, H. (1980). Stress, aging, and retirement. gsgrns1_s;
Hind_and_fiehayiors1_1. 93-110.

Sherman, L. K. G. (1980). The correlates of happiness in
post-separation adjustment(Doctora1 Dissertation,
University of Oregon. 1979L Dissertation_Ab§tracts

Intsrnstigns1‘_19,50228. (University Microfilms No.
80-05,800)

Sontag, M. s., Bubolz, M. M., 8 Slocum, A. C. (1979).
Perceived quality of life of Oakland County families: A
preliminary report.

Liying, Michigan State University Agricultural
Experiment Station.

Spreitzer, E., 8 Snyder, E. E.(1974). Correlates of life

satisfaction among the aged- Journal_of_§erontology.
22, 454-458.

137

Stock, W. A., Okun, M. A., Haring, M. J., 8 Witter, R. A.
(1983). Age and subjective well-being: A meta
analysis- In R- J Light(Ed ). Exalnation_stud1es1

Bsy1sg_snnss1, (Vol. 8, pp. 279- 301). Beverly Hills,
CA: sage.

Toseland, R., 8 Rasch, J.. (1979-1980). Correlates of life
satisfaction: An AID analysis. Ingsnnsg1sns1_gggzns1_sfi
As1ng_and_flunan_neyelonment. 10. 203-211.

Vancovering, V. G. R. (1974). An exploratory study of
middle-aged and older widows to investigate those
variables that differentiate high and low life
satisfaction (Doctoral Dissertation, Wayne State
University)- D1ssertat12n_Ahstracts_1nternationall_1s.
3895A. (University Microfilms No. 73-31,788)

Vinokur, A., 8 Selzer, M. L. (1975L Desirable vs.
undesirable life events: Their relationship to stress
and mental distress. J2urnal_of_£ersonality_and
25222212911_22121. 329-337-

Walker, R., Bubolz, M., 8 Lee, M. (1991). Contributions of
family resource management to quality of family life of
mid-life women and men. In J. Bauer (Ed.), £sm11y

n i W - e' ° t Ce u a e es
Co ' P d' 9 c
flszkshsp, Sponsored by Family Economics-Home Management
Section, American Home Economics Association Meeting,
Minneapolis, MN. 167-179.

Walker, R., Lee, M., 8 Bubolz, M. (1989). The effects of
family resources and demands on quality of life: A
rural-urban comparison of women in the middle years. In
H. L. Meadows 8 M. J. Sirgy (Eds. ), Qg_11§y;sfi_L1fis

.- e: 5 4. .- 1° ; 2:130‘u‘! ' - - . o: - !
th1rd_Qnal1ty:9f_narketing_c_nference- Blacksburq. VA:
Virbinia Politechnic Institute 8 State University, 397-
411.

Walker, R., Lee, M., Bubolz, M., 8 Keefe, D. (1990). Quality
of life of middle-aged women: Effects of family
resources and demands. In M. Carsky (Ed.), Erssssg1ngs

QQDEBE§I_IDEEI§§£§- Columbia, Mo: ACCI. 12-17.

Warr, P., Barter, J., 8 Brownbridge, G. (1983). On the
independence of negative and positive affect. lggrnsl

of_2ersonal1ty_and_§ocial_£sych21291111. 644-651.

Williams, T.R. 8 Bubolz, M.M. (1986). Family life events and
perceived stress: An examination of sex differences.
Unpublishedbpaper.

 

138

Williams, A. W., Ware, J. E., Jr., 8 Donald, C. A. (1981). A
model of mental health, life events, and social
supports applicable to general populations. Jsg:ns1_gg
W. 324-336-

Wright, S. (1960). Path coefficients and path regressions:
Alternative or complementary concepts? n1smsgg1ss1_1_:
189-202.

Yaar, E. (1976L Effects of socio-psychological variables on
subjective economic welfare. In B. Strumpel (Ed. ),

W Ann Arbor. MI: Insitute
for Social Research. 107- 130.

Zautra. A. J-. & Hempel. A- (1983)- Wine
W

C

e
W. Manuscript submitted
for publication, Arizona State University.

Zautra, A. J., 8 Reich, J. W. (1980). Positive life events
and reports on well-being: some useful distinctions.

MW. 657-670-

Zautra, A. J., 8 Reich, J. W. (1983). Life events and
perceptions of life quality: Developments in a two-
factor approach.

121-132.

"Triflii‘fliiflifliifliiiflig