. 1.3.11.3
.c at a.
$7?

.

.52.. T.

 

 

 

 

 

 

 

 

 

 

 

.
Q .35.
gist-ivy.
it
y .
it
(It =1

1m; x.

.2.
‘ is... a .
. ‘ua‘V-l-nfilfi . 1‘

...L£§..v. #..§......u....fi...

.71.: .39.. Juli
... .3. x is y .
XII-IR c)...\'1 In. 3?
”13:3...” nhrauwi‘. 9......
5... $39.. .fi-m.a
aw... .3 ii.

.

I‘
#4335. .

. t.
a. $13? .
‘fiabmmufih cit
i..2.........sr.u.1ufi. am.
3... nautifiqaafkii
. .....:.:... .5...

:

Lugsk.) .
a}. 1...
V. A .i .1:
E!!!..
. s. .
5! A 10:93.. 7 . 6.22..
his.u.?wrr.1§\drl3§!w«§i:7. .
, final. ‘1... .
s. ‘1‘
“1.5153113:
115‘”!- I v .
.. 9 33.3“». ...a....w}hanuw.%.
a k . a. .1 I!
‘0 if: . .1: (If

@flfiw ,.

Ht 1'.“ (:5

20b

LIDnnn I

Michigan State
‘ University

This is to certify that the
dissertation entitled

AGING IN JAPAN: IMPORTANCE OF SOCIAL
INTEGRATION

presented by
Kimiko Tanaka

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Sociology

 

 

  

We”? Professor’s Signal"re
W x; 2005/

Date

 

MSU is an afiinnatr've-action, equal-opportunity employer

so‘—-—.--.-I-'-I-n-o-t- .c...-_----o-.— _.

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K‘IProilAccatPresICIRCIDateDuemdd

AGING IN JAPAN: IMPORTANCE OF SOCIAL INTEGRATION
By

Kimiko Tanaka

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

DOCTOR OF PHILOSOPHY
Department of Sociology

2008

ABSTRACT
AGING IN JAPAN: IMPORTANCE OF SOCIAL INTEGRATION
By

Kimiko Tanaka

Intergenerational mutual care in Japan developed in the context of various factors,

including cultural ideals, centralization of the civil state, and the family unit called
the ie (X). Prior to the Meiji period, there were regional and class diversity in family

life and no cultural ideal for women to take care of the dependent elderly. However,
many scholars start with this patriarchal ie system (the system of primogeniture
legally recognized in the Meiji Civil Code in 1898) as the benchmark against which
to gauge the continuity, uniqueness, and change of eldercare during the twentieth
century. This has resulted in many scholars assuming that the cultural underpinnings
of the ie defined the care of ‘frail elderly’ as ‘women’s work’ undertaken to preserve
‘women’s morality.’ Since care has been pessimistically perceived as a caregiver’s
obligation, the elderly have too often been stereotyped as frail and dependent.

Owing to longevity and reduced disability among the elderly after the Second
World War, more elderly have challenged these stereotypes. Now perhaps Social
Gerontologists should start to view the Japanese elderly as proactively choosing their
opportunities to care and be cared for by important people beyond the family. This
new view has broadened studies of aging beyond just the maintenance of physical
and mental vitality to the social integration of older people with family, friends, and

communities.

In examining the relationship between social integration and the well-being of the
elderly, previous studies supported the role-enhancement perspective. For the elderly
who went through various life events, holding diverse roles is more likely to benefit
their well-being through enhancing individual resources and social connections.

The current study examines the importance of social integration on the well-being
of the Japanese elderly using two waves of the Nihon University Japanese
Longitudinal Survey of Aging. The findings support the role-enhancement
perspective. Especially, residence showed complex effects. The rural elderly had
greater odds than the urban elderly of having a disability. However, the rural elderly
had lower odds than their urban peers of feeling depressed. The advantage of
ruralites over urbanites in escaping from depression may arise from a greater

integration of rural people into social networks of mutual care.

Copyright by
Kimiko Tanaka

2008

ACKNOWLEDGEMENTS

My deepest gratitude goes to Nan E. Johnson, Ph.D., who, as my chair and
mentor, generously gave her knowledge, time, and encouragement in assisting my
study at Michigan State University. I am grateful to my committee members, Rita
Gallin, Khalida Zaki, and Lori Post, for their insight and encouragement. I also
would like to express my gratitude to all professors I met at Michigan State
University for enabling my academic life and making it productive and
unforgettable.

This study used data from the Nihon University Japanese Longitudinal Study of
Aging (NUJLSOA) conducted by the Nihon University Center for Information
Networking. I am especially grateful to the principal investigator, Yasuhiko Saito,
Ph.D., for his kind and generous support in granting me access to the two waves. I
also would like to thank the Graduate School at Michigan State University for
providing me Dissertation Completion Fellowship. It greatly helped me to work on
my dissertation.

Last not but not least, many thanks to my husband, my family, and my friends for
their tremendous patience and support, and to David Barr and Elizabeth Barr for
their encouragement teaching me academic English. I hope this study will help to
increase understanding of aging and health in Japan as well as other nations in the

world.

TABLE OF CONTENTS

LIST OF TABLES ........................................................................... viii
LIST OF FIGURES ............................................................................ x
CHAPTER 1. INTRODUCTION ............................................................ 1
CHAPTERZ. BACKGROUND

The Shifiing Roles of Women in lntergenerational Mutual Caregiving in Japan:

The Importance of Peace, Population Growth, and Economic Expansion ............. 7
Introduction ................................................................................. 7
The Establishment of the Ritsuryt") State (645 AD. — 900 AD.) .................. 8
Moving From the Ritsuryo State to the Tokugawa Period
(1603 AD. - 1868 AD.) — Birth of the ie ............................................ 13
The Tokugawa-Period (1603 AD. — 1868 AD.) — Transition
to Modernization ......................................................................... 17
The Meiji Period to the Second World War (1868 AD — 1939 A.D.):
Modernization of the State and the [e System ....................................... 29
Postwar Japan — The Path to Becoming the Most Long-Lived Nation
in the World .............................................................................. 35
Conclusion ................................................................................ 42

CHAPTER 3. THEORETICAL FRAMEWORK
Aging and Family in Japan in the Twenty-First Century: Social Recognition

of the Importance of Social Integration on Health ........................................ 44
Introduction ................................................................................ 44
Why Social Integration is Important in Studying Aging? ................................. 44
Theoretical Framework: Understanding Social Integration
— Multiple Role Theories ............................................................... 53
Social Integration and Health of the Elderly: Role Enhancement
or Role Strain ............................................................................. 56

vi

Incorporating the Life Course Framework to the Study of Social

Integration and Health .................................................................. 62
Statement of Questions .................................................................. 68
CHAPTER 4. METHODOLOGY .......................................................... 71
Introduction ............................................................................... 71
Sample Description of the Nihon University Japanese Longitudinal
Study of Aging ........................................................................... 72
Method of Sampling ..................................................................... 74
Measures .................................................................................. 77
Statistical Procedures .................................................................... 89

CHAPTER 5. RESULT (1)

Cross-Sectional Analyses of the First Wave— Research Question 1 and 2 ............ 91
Research Question 1 ..................................................................... 92
Research Question 2 ................................................................... 101
Summary ................................................................................ 110

CHAPTER 6. RESULT (2)

Longitudinal Analyses of NUJLSOA Data Set — Research Question 3 113
Research Question 3 ................................................................... 115
Summary ................................................................................ 125

CHAPTER 7: Conclusion

Aging in Japan in the Twenty-First Century ............................................ 129
Summary of My Dissertation ......................................................... 129
Policy Implications .................................................................... 135
Implications for Future Studies ...................................................... 139

APPENDICES ............................................................................... 145
Figures and Tables ..................................................................... 145
Japanese Vocabulary .................................................................. 164

BIBLIOGRAPHY ........................................................................... 166

vii

Table 1.

Table 2.

Table 3.

Table 4.

Table 5-A.

Table S-B.

Table 6-A

Table 6-B

LIST OF TABLES

Paired-Sample t-tests of dependent variables .............................. 151

Multinomial Logistic Regression Comparing (1) Respondents

with one or more disability on ADL or IADL scales, (2)

Respondents with at least one of the answer on ADL and IADL

scale missing, and (3) Respondents with no physical disability
(Reference Category) .......................................................................... 152

Multinomial Logistic Regression Comparing (1) Respondents

who rated their health better than average, (2) Respondents who

rated their health worse than average, and (3) Respondents who

rated their health average (Reference Category) .......................... 153

Multinomial Logistic Regression Comparing (1) Respondents

with one or more scores on depression scale, and (2) Respondents

with at least one ore more missing scores on depression scale,

and (3) Respondents with zero score on depression scale (Reference
Category) ....................................................................... 154

Multinomial Logistic Regression Models Comparing

(1) Respondents with one or more disabilities on ADL or IADL

scales and (2) Respondents with no physical disability (Reference
Category) ...................................................................... 155

Multinomial Logistic Regression Models Comparing

(1) Respondents with at least one of the answer on ADL and IADL
scales missing and (2) Respondents with no physical disability
(Reference Category) ......................................................... 156

Multinomial Logistic Regression Models Comparing

(1) Respondents who rated their health better than average and

(2) Respondents who rated their health average (Reference

Category) ...................................................................... 157

Multinomial Logistic Regression Models Comparing

(1) Respondents who rated their health worse than average and

(2) Respondents who rated their health average (Reference

Category) ...................................................................... 158

viii

Table 7-A

Table 7-B

Table 8

Table 9

Table 10

Multinomial Logistic Regression Models Comparing

(1) Respondents with one or more scores on depression

scale and (2) Respondents with zero score on depression

scale (Reference Category) ................................................... 159

Multinomial Logistic Regression Models Comparing

( 1) Respondents with at least one or more missing scores on
depression scale and (2) Respondents with zero score on

depression scale (Reference Category) ...................................... 160

Multinomial Logistic Regression Models Comparing

(1) Respondents had at least one disability in wave 1 but zero
disability in wave 2, (2) Respondents had zero disability in

wave 1, but at least one disability in wave 2, (3) Respondents

having at least one disability on both wave 1 and wave 2, (4)
Respondents with at least one of the answer on ADL and IADL

scales missing, and (5) Respondents having zero disability on

wave 1 and wave 2 (Reference group) .................................... 161

Multinomial Logistic Regression Models Comparing

(1) Respondents rated their health average in wave 1 and

wave 2, (2) Respondents rated their health below average

in wave 1 and wave 2, (3) Respondents who rated their health

better in wave 2, (4) Respondents who rated health better in

wave 1, and (5) Respondents whose answer of self-rated health

was missing, and (6) Respondents rated their health better than

average in wave 1 and wave 2 (Reference Group) ....................... 162

Multinomial Logistic Regression Models Comparing (1)

Respondents with one or more scores on depression scale in

wave 1, but zero score in wave 2, (2) Respondents with zero

score on depression scale in wave 1, but one or more scores

in wave 2, (3) Respondents with one or more scores on

depression scale in both wave 1 and wave 2, (4) Respondents

with one or more scores on depression scale missing,

and (5) Respondents with zero score on depression scale in

both wave 1 and wave (Reference group) ................................. 163

ix

LIST OF FIGURES

Figure 1. Map of Japan .................................................................. 146
Figure 2. Population of Japan .......................................................... 147
Figure 3. Total Fertility Rate (TFR), 1925 to 2004 148
Figure 4. Labor Force Participation Rate by Sex ................................... 149

Figure 5. Life Expectancy at Birth for Japanese Men and Women ................ 150

CHAPTER 1

INTRODUCTION

Japan became the nation with the greatest longevity in the world in 2005, having
the longest life expectancy at birth (82 years in 2006), the largest portion of the
elderly (25 percent), and the smallest proportion of the population 15 years and
younger (13.6 percent) (Ota 2006). This resulted in an urgent agenda for promoting
healthy aging in Japan. In this context, research that identifies factors contributing to
healthy aging in Japan became important to advance sociological theory about aging
as well as to guide policy initiatives to promote healthy aging in Japan and other
nations in the world.

In the past, studies of aging in the US. and Japan explicitly or implicitly assumed
that the elderly were dependent and passive recipients of care. Elders and children
have been perceived as socially marginal and incapable of engaging in relationships
beyond their families (Hazan 1992; Thang 2000). In the United States, the elderly
have too often been associated with problems rather than promises, although the idea
that older persons cannot maintain active participation in society and the economy
appears to be more myth than reality (Cohen 2005; Cohler and Altergott 1995;
Garstka, Shaner and Strahrn 1994; Palmore 2005; Robertson 1995; Troll 1995). Such
an “elderly myth” treated the elderly as if they were a homogeneous group marked

by stereotypes, which Garstka et a1. (1994) reduced to four negative notions

(severely impaired, shrew/curmudgeon, recluse, despondent) and three positive
stereotypes (perfect grandparent, John Wayne conservative, golden-ager).

In Japan, elders and children have also been perceived as socially marginal,
belonging to the supernatural world with God (Thang 2000). Benedict (1946)
described the life course of Japanese people as shaped like a U-curve, with
maximum indulgence and freedom allowed to babies and the elderly. The ethics of
Confucianism taught children to respect the elderly because not only were they
perceived as knowledgeable, but also they were described as frail and dependent. For
example, one of the proverbs in Meiji Iroha Karuta (flfiifilfl 611110613, a sort of
traditional and popular Japanese card game utilizing proverbs, says, Oitewa k0 m'

shitagae (%L\111¥l:17£i). It means that, when you are old, obey your children.

The ideology behind this proverb derived from Confucian precepts for women that,
“When you are young, obey your parents. When you marry, obey your husband.
When you are old, obey your children.” This precept suggests that stereotypes
toward the elderly not only homogenized as dependents, but also gendered the life
course of the elderly. In both contexts, “care” meant the caregiver’s responsibility
and obligation of looking after someone weak and dependent, and care receivers
were perceived as passive recipients.

Today, owing to longevity and reduced disability among the elderly, more elderly
are coming to be seen as proactively choosing their opportunities to care and to be
cared for by important peOple in their later life course. In the US, where individual

choices and autonomy are highly valued and preferred (Hashimoto 2000; Long and

Harris 2000), many elderly actively sustain or develop social networks to fulfill their
life, to make their life meaningful, and to stay connected to society. As Klinenberg
(2002) observed, longevity is shifting its meaning from individuals simply living
longer to “forging new opportunities for creating things, for making or developing
meaningful relationships, for contributing to society, to family, and to friends” (p
40).

In Japan, in its history of nationalization, modernization, and industrialization, the
responsibility and obligation of the eldest son to support his elderly parents was once

legitimized by the stem family system known as ie seido (INK) (Hamabata 1990;

Ochiai 1996; Thang 2000). In other words, Japanese elderly were legitimized to be
dependent on their family members. However, after the new Civil Code in 1946
replaced the ie seido with the conjugal family system, industrialization,
democratization, and urbanization increased the mobility of younger generations and
contributed to the nuclearization of the family (Thang 2000). Reflecting such
dramatic institutional change, increasingly the elderly started to show their
preference not to be solely dependent on family members for care as passive
recipients (Ogawa and Retherford 1997). According to Ogawa and Retherford
(1997), expectations of old-age support from one's own children declined from 65
percent in 1950 to 13 percent in 1996. Although the government remains criticized
for heavily relying on families for the filial care for the elderly, there are

improvements in the quantity and quality of private and public care facilities as well

as various activities and volunteering programs for the elderly provided by local
governments and non-profit organizations.

In the twenty-first century, these facilities, programs, and activities aim to provide
the elderly opportunities to avoid social isolation in the later life course, and to

remain in the circle of harmonious integration (wa: HI), where people can ask for
reliance on the indulgence or the goodwill of one another (amae: Hi) beyond the
traditional stem family (ie: I) in Japan. To be a good elderly person in Japan, one

needs to be a “socially engaged individual who is involved in activities that
incorporate social interaction; failure to maintain activity and social involvement
invites loss of well-being” (Traphagan, 2004: p. 9). In such a context, the importance
of social integration on the well-being of the elderly was recognized.

Advancing research on social integration and health in later life not only has
important policy implications for paths to healthy aging, but also calls for new
research to extend our knowledge of what social factors are important. The research
on social integration and health has thus emerged as an urgent and important issue of
both public and scientific concern. In the twentieth-first century, as more nations in
the world continue to have expanding populations of the elderly, scholars are in the
midst of a theoretical revision to explain the relationship between social integration
and health in later life. The purpose, as well as the major scientific merit, of my
dissertation research is to advance the theoretical and empirical study of social

integration and health by examining a longitudinal survey of the elderly in Japan.

The broader merit of this study is to identify and develop recommendations to
promote healthy aging.

The next chapter looks at historical changes in the cultural superstructure defining
the proper organization of eldercare. It provides historical evidence to explain how
intergenerational mutual care in Japan developed in a context of various factors,
including cultural ideals, centralization of the civil state, and the family unit called
the ie. It shows that the cultural superstructure defining the proper organization of
eldercare is not immutable. Rather, it has changed historically because of its
interaction with period-specific demographic, economic, and political realities. This
historical background is a key to reconsider the social organization of eldercare in
contemporary Japanese society, the most long-lived society in the world.

The third chapter describes the theoretical framework and reviews previous
literature regarding social integration and health. Previous studies of the role
enhancement perspective suggest that having multiple roles is linked to better health
by preventing social isolation, increasing social connectedness to the society, and
buffering stress caused by role loss or role strain. Incorporating the life course
framework, I argue that the elderly who occupy not only many but also diverse roles
over time may be buffered from role loss or may have enhanced social support in
later life and thus have better emotional and physical outcomes than those who do
not. I also argue that these effects are expected to differ by gender, residence, and

birth cohort. This chapter ends with three research questions, based on the theoretical

framework adopted and previous studies, which are answerable from a national level
longitudinal survey (wave 1 and wave 2) of elderly Japanese people.

The fourth chapter first describes the data, two waves of the Nihon University
Japanese Longitudinal Survey of Aging (NUJLSOA). Then I explain how variables
were measured and used in analyses and describe why the statistical method I chose
was the proper method. The fifth and the sixth chapters report findings from
statistical analyses. The fifih chapter includes findings from the cross-sectional
analyses of the first wave of the NUJLSOA data set, and the sixth chapter includes
findings from longitudinal analyses looking at changes in health conditions of the
elderly respondents between the first wave and the second wave of the NUJSOA data
set.

Finally, the last chapter summarizes my dissertation research as well as addresses
policy implications and future research based on the findings and limitations of my

dissertation.

Chapter 2

BACKGROUND
The Shifting Roles of Women in lntergenerational Mutual Caregiving in Japan:

The Importance of Peace, Population Growth, and Economic Expansionl

Introduction

Various scholars have discussed aging in Japan as a result of demographic,
cultural, and economic transformations after World War II that altered the balance of
intergenerational caregiving (Asai and Kameoka 2005; Maeda 2004; Ogawa and
Retherford 1993; Retherford, Ogawa and Sakamoto 1996). The major problem in
some of these literatures is the historical starting point scholars choose, because they
usually assume that intergenerational caregiving changed only after that point. For

instance, many scholars start with the patriarchal ie system, ie seido (X‘ME), the
system of primogeniture legally recognized in the Meiji Civil Code in 1898, as the
benchmark against which to gauge the continuity, uniqueness, and change in elder
care during the twentieth century (Asai and Kameoka 2005; Long and Harris 2000).
These scholars assume that the cultural underpinnings of the ie (2), pronounced
"ee-eh", define the care of ‘frail elderly’ as ‘women’s work’, undertaken to preserve

‘women’s morality’. This assumption is misleading because the ie did not clearly

 

' This chapter was presented at the annual meetings of American Sociological Association, New
York, August 12, 2007.

assign elder care to the wife of the eldest son until the Meiji period (1868 AD. —
1912 A.D.).

The goal of this chapter is to show that the cultural superstructure defining the
proper organization of elder care is not immutable. Rather, it has changed
historically because of its interaction with period-specific demographic, economic,
and political realities. This historical background is a key to understanding the social
organization of elder care in contemporary Japanese society, the most long-lived
society in the world. Towards this goal, I have organized this chapter into five
historical periods according to the predominant mode of government: (1) the
establishment of the Ritsuryc‘) State (645 AD — 900 AD), (2) the transition from the
Ritsuryo State to the Tokugawa Period, (3) the Tokugawa Period (1603 AD - 1868
AD), (4) the Meiji Period until the Second World War (1868 AD - 1939 AD), and

(5) postwar Japan.

The Establishment of the Ritsugyo State (645 AD. - 900 AD.)
During the Yayoi era (200 BC. — AD. 300), the Japanese people experienced two

innovations — bronze and iron articles introduced from mainland Asia, and settled
agriculture, especially wet-rice cultivation. These improved people’s lives compared
to the preceding stone age, called the Jomon era (10,000 - 200 BC). The Jomon era
is known for pottery that bears cord markings, which reflect cultural enrichment
prior to this era. Prior to the Jémon era, 10,000 to 7,000 years ago, people’s lives

were dependent on fishing and hunting. They lived in partly sunken, pole-and-thatch

dwellings in scattered and small arboreal communities near the seashore (Totrnan
1982). They obtained fish by netting, diving, digging, and line fishing, and they
supplemented this fare with game and forest fruit, which continued until around the
third century BC. (Totman 1982). During the Jomon era, the single-room house was
enlarged and its framing was reorganized so that they no longer needed a center pole
to support the roof, and bare earth was replaced by drier stone flooring (Totrnan
1982)

Beginning about the third century BC, by replacing and supplementing the
hunting, fishing, and gathering habits of previous generations, settled cultivation
contributed to the population growth in the Yayoi period (Farris 1985). Increased
food supply through farming enabled people to live in much larger villages, some
estimated to include 500-600 houses, which gradually required and enabled trading
activities, a hierarchical community, and political consolidation (Totrnan 1982).
Consequently, social inequalities widened as the nation moved from a system based
on unstable slash-and-burn agriculture to more productive settled agriculture
accompanied by population growth. A more reliable food supply probably spurred
population increase by depressing the mortality rate. Before the Yayoi era, the
population was estimated to be between 120,000 and 350,000 people. By the end of
the Yayoi period, iron farming tools and more sophisticated irrigation techniques had
contributed to further population growth, increasing the population to between 1.5

million and 4.5 million (Farris 1985). Despite archaeological evidence including

burial practices, the lack of reliable data makes the trends fiom 300 AD. to 645 AD.
rough estimates (Farris 1985).

According to a Chinese report of ca. AD. 297, Chinese visitors observed the
Japanese people (“the people of Wa“) living in houses where father and mother,
elder and younger, slept separately. They also reported “men of importance” enjoyed
four or five wives and the deference of their inferiors. They also described, however,
that there was no distinction between father and son, or men and women, in their
meetings and their deportment (Totman 1982). Even with very limited resources, it is
possible to infer that the system of primogeniture was not established before the
Ritsuryo state.

In the mid-seventh century, emerging powers in East Asia, especially China’s
Tang empire, became a growing threat to Japanese rulers. In response, Japanese
political leaders established the Ritsru'yo state (645 AD. — 900 AD.) — a Chinese-
style centralized civil state based on legal codes. The emperor Tenji proclaimed that
the state bureaucracy was to be a rationalized and systematic organization, and
announced a plan to use Chinese systems of taxation, local government, and land
tenure in the Taika Reform Edict in 646 AD. Under Tenji’s hegemony, the Japanese
people were enrolled in a civil register, and surnames were established for the
purposes of taxation and conscription. The first census followed in 670 AD. (Farris
1985). Errors of coverage and accuracy, however, make these registers unreliable as
a source of population analysis. Despite these difficulties and limitations, Farris

(1985) used stable population analysis and estimated that the life expectancy at birth

10

ranged from 28 to 33 years. The infant mortality rate was so high that, although
peasants were given grants of rice paddies at birth in 690 AD, children aged five or
younger in calendar years soon became the exception (Farris 1985).

The Ritsuryo state brought the ancient Chou principle of inheritance by the eldest
son of the sole legal wife, at least among the leading nobility (Holcombe 1997).
Because the Ritsuryo’ state also brought social stratification based on class, age, and
gender, this resulted in official resources containing very little evidence about
women, although they played a surprisingly powerful role as toji (7] E). To promote
Chinese concepts of patrilineality and male dominance, official documents recorded
conventional gender-biased images of the productive activities of men (Yoshie
2005). According to hidden sources such as old tales (setsuwa: BEE) relating the
lives of common people and to archaeological findings such as wooden tablets
(mokkan: 7km) used for keeping routine records when paper was an expensive
commodity, toji played a major role in rural society, managing agricultural
enterprises and supervising labor (Yoshie 2005). Thus, although toji became the term
“housewife" (shufu: $95) in later history (Goto 2002), toji should not be regarded as
“wives’ who exercised leadership as proxies for their husbands in the local
community (Yoshie 2005). On a higher level, 6t0ji (7(7) 5), grand toji, managed
productive enterprises within their independent residences. In addition, from
studying lineage records and inheritance patterns, historians have concluded that

Japan was not organized along patrilineal lines; it was a bilateral society in which

lineages of both father and mother were important (Yoshie 2005). Hence, there were

11

many gaps between the idealized patriarchal Ritsuryo state and actual social
conditions. Efforts at both central and local levels to cope with these gaps gradually
transformed the nation to the Ritsuryo system, which better fit the social and
economic actuality of the time (Yoshie 2005).

Along with a Chinese-style centralized state, the Ritsuryo system also brought the
Confucian spirit of filial piety to Japan (Kasugai 2004). For instance, the Ritsuryé’
system required men to pay a tax depending on their age. Men ages 61-65 years paid
only half of the tax paid by men between 22 and 60 years old; and men ages 66 or
older paid no tax (Kasugai 2004). Because of the low survival rates of infants and the
elderly, their existence was ofien described as spiritual, which is reflected in the
saying, “Oite futatabi chigo m’ naru” (tb‘fllilfllztié). It means that,
people become a child again as they become the old (Kasugai 2004).

In such a context, senile dementia was considered an existence that moved a
person close to God, and people warmly watched over them. From pictures on scrolls
that depict children holding the hand of the elderly and the elderly walking while
holding children’s shoulders (Kasugai 2004), it appears that children, rather than
daughters-in-law, took care of the elderly. Although Confucianism taught children to
respect the elderly, the cultural ideal of a woman, a daughter-in-law, taking care of

the elderly was not established. Many picture scrolls (emaki: fig) and old tales

(setsuwa) also described the elderly farming and taking care of grandchildren
(Kasugai 2004). The elderly also were described as great contributors to the society

by virtue of their wisdom and experience. Along with women and children, however,

12

the elderly were also described as socially weak and considered to be socially
marginal. This was reflected in old tales in which the young left the elderly in the

mountains (ubasute-yama: #151 LIJ). These tales appear, however, to exist for

people to criticize such impious children; hence, it appears that children took
caregiving roles of the elderly (Kasugai 2004). Due to high infant and child
mortality, there were also old tales that described the elderly worrying and grieving
about their care in later life (Kasugai 2004). In other words, women’s morality to be
responsible for elder care was not prevalent, not even universalized. It appears that
morality was contextual rather than universal. Rather than being regulated by a
centralized government with formal and abstract rules and orders, moral life was

preserved in a community, especially among peasants.

Moving From the Ritsugyo State to the Tokugawa Period (1603 AD. - 1868

AD.) - Birth of the ie
During the Classical Period (710 AD. —- 1185 AD), at least among the elite, the

estate was equally divided among all the children, including daughters (Smith 2000).
From the tenth century to the eleventh century, gradually from upper noble classes,
the concept of ie emerged; and by the beginning of the Tokugawa period, the concept
had developed to be systematized as the ie system, a way to bequeath government
posts from fathers to sons through primogeniture. 1e is a term that is widely defined
as the Japanese "stem family" in current literatures (Hashimoto 1996; Kawano 2003;

Knight and Traphagan 2003; Ochiai 1996); however, as various anthropologists

13

argued, originally the term ie connoted a “household” based on coresidence that
could include members unrelated by blood, marriage, or adoption. In other words,
the ie was a term used to describe traditional Japanese “households” based on
participation in the ie economy and coresidence, rather than on kinship ties or the
Confucian dyad of father and son.

Conceptually, the ie included not only its members but also the ie economy (e. g.,
crafts production, farming, retail sales) and resources (e.g., tools and land) (Smith
2000). The ie was the basic economic unit in society. Rather than biological
continuity, the heart of the ie was to continue the ie throughout generations (N akane
1967, 1970). The core structure of the ie was not the household head and his spouse,
and it was not narrowly defined as the idealized Confucian dyad of father and son.
To continue the ie economy, adopted men of proved competence could inherit the ie,
occasionally at the expense of direct biological heirs (Smith 2000). Thus the ie could
stand in contrast to the kinship-based term kazoku (gfi), a term invented in the
nineteenth century to deal the Western idea of the nuclear family in Japan (Johnson
1964)

As the militarization of Japanese society flourished under the samurai class
during the Medieval Period (1185 AD. -1600 AD), the social position of women
declined. Many women lost their right to inherit property and to exercise public
authority. The political leadership fostered equal inheritance among sons during the
Kamakura Period (1185 AD. — 1336 AD.) to prevent families from building an

economic power base (Smith 2000). Although some variations (inheritance by the

14

firstbom female or last-born child) were seen across regions during the Muromachi
period (1336 AD. -1573 AD), inheritance through primogeniture gradually
replaced equal inheritance among sons (Smith 2000). Bequeathing the authority and
resources to the eldest son increased the chance that the ie estate could be passed
down and maintained throughout generations.

Such an ie system was not prevalent except among noble families and samurai
families until the middle of the nineteenth centrrry (the Tokugawa period) (Benedict
1946). In fact, marriage was an official ceremony only for the privileged classes

(Kasugai 2004). Around the tenth century, men could have multiple wives and

mistresses; but noble men were restricted to one lawful wife (seisai: IE!) during

the eleventh century (Goto 2002). Wives of the upper classes took care of the
household not only by preparing meals but also by providing maintenance of
weapons; managing the food supply for the ie; managing apprentices, followers, and
their families who supported the ie; and managing the ie economy, as well as the
marriage to sustain and nurture the ie. Afier the death of their husband, these wives

became goke (fig), who often had the final say about the management of the ie

(Got?) 2002). Therefore, the division of labor was not strictly and clearly gendered.
Both husbands and wives supported each other to nurture the ie by increasing their
land and resources through wars and political contributions, by increasing their
followers through taking care of their families to gain their trust, and by expanding

their land and resources through a marriage of convenience (Goto 2002). The

15

important status of wives as decision makers of the household economy was also the
same for other lower classes (Goto 2002).

Prior to the Tokugawa period, most of the population was rural, and the economy
was based on subsistence agriculture due to the lack of an urban population (Hayami
and Kurosu 2001). Due to high mortality and unstable socioeconomic conditions
caused by wars and famine, few elderly could be taken care of by their relatives.
Even when elders in the lower classes could survive, it appears that many of them
lacked immediate kin and lived on a lord’s manor in a community as servants
(Kasugai 2004). It appears that ideal caregivers were relatives, not strictly women

(Kasugai 2004). One of the Buddhist tales, Shasekishfi (seam, talked about a

priest who suffered from paralysis mourning his situation that he did not have any
wives or children to take care of him. The elderly in noble families and samurai
families could enjoy their authority as the succession of the ie became important.
Some upper-class elderly, however, decided not to be dependent on the ie and spent
their later life in temples as Buddhist priests or nuns (Kasugai 2004). Hence,
although the elderly were scarce in the past but are abundant in the present, there was
diversity in family and aging before the Tokugawa period. Although most did not
survive to old age, some worried about their care in later life, some tactically used
their power to sustain autonomy in later life, some detached from the ie by becoming
priests and nuns, some wished to be dependent on their children, and others

contributed to society as laborers or caregivers of grandchildren.

16

The Toku awa-Period 1603 A.D. — 1868 A.D. - Transition to Modernization

   

During the Tokugawa period, the Tokugawa shogunate ordered the populace to
register at a Buddhist temple and practice ancestor worship (Mori 1993). The main
purpose of this order was to suppress Christianity (Cornell 1996; Kawano 2003; Kito
2000; Yasuda 1992), which places one’s relationship to God before all others; hence,
it posed a threat to the subjects’ loyalty to the feudal lords and the shogunate
(Kawano 2005).

Temples began to function as centers of religious and household registration

(shumon aratame-ché: BREW“) (Hanley 1972; Hayami and Kurosu 2001). The

government used the feudal structure of the ie as a fundamental social group to
regulate and govern people’s behaviors and ideas. Nakane (1970) claimed that the
effectiveness of regulations reaching every village and household even up in the
remote hill areas was not simply a reflection of the power of the Tokugawa
shogunate, but also a by-product of the layers of loyalty and the lord-vassal

relationship (oyabun and kobun: EQTQ‘). According to Nakane (1970), China had

a strong horizontal social organization, and their central administration was
constructed on horizontal bases, such as the network of patrilineal clan organizations,
guilds and castes, had difficulty in extending the basis of its authority throughout the
entire population. Hence, Japanese vertical social organizational pattern — Japanese
cultural emphasis on putting the collective over the individual to preserve
harmonious integration (wa) in families and communities —- was natively rooted

before the advent of Chinese influence.

17

Demographically, the Tokugawa period was innovative compared with the
preceding period. The religious investigation registers or the village population
registers (shumon aratame—cho') were census-type listings of village populations
compiled every year, with information added in many cases about births, deaths, and
marriages that occurred during the year (Saito 1992). Significant improvement in
historical demography to use such registers contributed to current scholars revising
previous studies of the relationship between demography and economy in the
Tokugawa period.

About 1730, Japan ranked in size with France, holding a population of between
20 and 30 million. A century and a half later, it fell to a lower ranking with Italy,
considerably smaller than France and Germany. It took Japan close to 200 years to
double its population level to reach 60 million (Cornell 1996). Hence, the first
generation of population historians up to the early 19608 pointed to a stagnating
population trend during the Tokugawa period due to poor economic conditions of
both samurai and peasant classes (Hanley 1972).

The literature published in the past fifty years in the field of Tokugawa historical
demography analyzed that population stagnation during the Tokugawa period could

be explained by the Malthusian notions of “preventive checks” such as abortion and

infanticide (mabiki: MEI 3) and “positive checks” such as famine and epidemics

(Jannetta and Preston 1998; Saito 1992). These literatures analyzed that fertility was
high due to high infant and maternal mortality, and lack of effective contraceptive

knowledge and skills. These factors led the Japanese to practice abortion and

18

infanticide when too many children were conceived or born (Kito 2001). Similarly,
these literatures argue that population stagnation is consistent with the Marxist
notion of exploitation: both positive and preventive checks to ameliorate starvation
operated directly to kill peasants who were exploited by samurai lords (Cornell
1996)

In the past 20 years, new evidence based on the religious investigation registers or
the village population registers have refuted such Malthusian and Marxist
assumptions of population stagnation during the Tokugawa period, and these suggest
that the majority of peasants enjoyed rising productivity and a higher standard of
living (Hayami and Kurosu 2001). Scholars argued that establishment of the
Tokugawa shogunate led to a flom'ishing of castle towns and post towns, and rural
residents were forced to increase the efficiency of agricultural productions to catch
up with grong urban demand (Hayami and Kurosu 2001). In addition, the spread
of commercial and handicraft industries enabled an increasing number of people to
make a living outside of farming because they could leave for commercial centers
(Hanley 1972). According to Hanley (1972), economic development in the
Tokugawa period can be likened to that in England in the century prior to the
Industrial Revolution. The English limited family size because they understood that
too many children could threaten their ability to accumulate resources. For instance,
the village of Fujito experienced economic development but limited families by
postponing the age at marriage for women to their mid-thirties, and sending younger

sons out to work in another village or to be adopted in by another household (Hanley

19

1972). Hence, the new evidence suggests that, prior to industrialization, although
famine and disease contributed to slow population growth (and there were regional
differences), Japan experienced increasing urbanization, a growth in the commercial
sector, rising incomes, and gradual commercialization of agriculture, all good
indicators of a growing economy (Cornell 1996; Hanley 1972; Hayami and Kurosu
2001).

Looking at fertility in detail, the well-documented village of Yokouchi reported in
the religious investigation registers or the village population registers that for women
born before 1700, the Total Fertility Rate (TF R), defined as the average number of
children a woman would have, assuming that current age-specific birth rates
remained constant (Weeks 2002), was 5.8 children per woman. For the 1701-1750,
1751-1800, and post-1800 periods, TFR decreased to 4.0, then to 3.2 and 3.4
respectively. The TFR was low due to unregistered infant deaths. Infant deaths
before the next census-taking date were likely to be under-recorded. Taking this
deficiency into account, the decrease in TFR was from about 7 to 4—5 (Saito 1992).
This reduction of TFR was also accompanied by the rise in life expectancy. For the
village of Yokouchi, the life expectancy at 2 years increased from 33 to 43 years
(Saito 1992).

Patterns of marriage and fertility were geographically diverse in the Tokugawa
period. Hayami and Kurosu (2001) divided Japan (excluding Hokkaido and
Okinawa) into three regions (Figure 1) -— the northeast, central, and southwest -

where necessary historical sources were available, and analyzed regional diversity in

20

demographic and family patterns based on the data that are samples from within the
larger region. According to Hayami and Kurosu (2001), the northeastern region was
characterized by an extremely low median age at marriage (19 years old for men and
13 years old for women). For women, this age reached 17.1 years by the end of the
period. In this region, women also stopped having children early (about 33 years),
and the number of births per couple was quite low despite early marriages (3 to 4
children per couple) (Hayami and Kurosu 2001). Another characteristic of this
region was that people in this northeastern region temporarily migrated to other
regions in Japan, often cities, to supplement their income during the winter

(dekasegi: 1:11???) Although Japan was closed to international immigration and

emigration for more than 200 years (1641 A.D. — 1853 AD), there were both
temporary and permanent internal migrations for purposes of employment (dekasegi)
(Cornell 1996). Because people migrated to other places for income on a temporary
basis, marriage was a way to keep both men and women bound to villages and
households (Hayami and Kurosu 2001). The word dekasegi implies that they would
come home after a while, reflecting parental power over their children (Saito 2000b).
These temporary internal migrations in this region lessened the risk of pregnancy and
lengthened the birth interval, which in turn reduced the overall number of births
(Hayami and Kurosu 2001). In this region, Saito (1992) found that peasant women
also had to limit births by practicing infanticide due to poor economic conditions, but

this was not because they had too many children.

21

In the central provinces (Figure 1), where relatively large preindustrial cities were
located, average age at the first marriage was higher than in the northeastern
provinces (about 28 years for men, and 20.5 for women, in the late eighteenth
century and the mid-nineteenth century). The number of children per couple was also
higher than the northeastern provinces (5.9 to 6.5 children per couple). Although
some people in this region were work migrants, they migrated on a permanent basis,
rather than on a temporary basis (Hayami and Kurosu 2001). Lower-class people
were more likely to migrate to other places of Japan to supplement their income
(dekasegi), and in contrast to the northeastern region, the majority of them migrated
to other places in Japan, often cities, to supplement their income prior to marriage.
Therefore, many families in the lower stratum risked the termination of the lineage

(zekke: fig), but families in upper stratum married early and tended to have more

children to maintain the family line (Hayami and Kurosu 2001).

Unlike the central regions, the southwestern regions (Figure 1) had no large cities;
but the age at marriage was high, just as in the central regions. Marital fertility rate
and the number of premarital births were high; and divorce, remarriage, and
premarital childbearing were common in this region (Hayami and Kurosu 2001). The
higher standards of living enabled peasants in southwestern Japan to limit their
family size not because of poverty, but as means of adjusting the size of the family
labor force to changing economic environments (Saito 1992). Hence, there was

regional and class diversity in marriage and fertility in the Tokugawa period.

22

Historical demographers and economists have argued that poor economic
conditions caused many villages to limit their family sizes and, thus, for population
size to stagnate during the Tokugawa period (Taeuber 1958). But other scholars have
claimed that peasants limited their family size, even resorting to infanticide, to
maintain a favorable man-to-land ratio and to attain a higher standard of living
(Hanley 1972; Hayami and Kurosu 2001; Saito 1992). Taking a different
perspective, Cornell (1996) argued that, rather than using deliberate controls on
fertility, people controlled their family size by natural methods, such as through
long-term breast-feeding practices and spousal separation through migration to
emerging and developing cities. The fertility decline in the Tokugawa Period refutes
the underlying assumption of the demographic transition theory that people were
ignorant about fertility control in preindustrial societies. Historical demography is
still in its development in Japan, and more research is necessary to interpret patterns
of marriage and fertility for both men and women during the Tokugawa period.

Mortality was high during the Tokugawa period. The life expectancy at birth was
estimated to be in the lower thirties from 1776 to 1875 (Cornell 1996). Bad weather,
poor harvests, high prices, and possible famine were prevalent in modern agriculture
when vaccination and medical treatment were not available. There were major food
crises in the 17308, 17808, and 18308. The Japanese population was large enough to
have epidemic and infectious diseases as well. Epidemics such as smallpox, measles,
influenza, and diarrheal diseases also raised death rates (Cornell 1996). In addition,

infant mortality rates were very high, implying one death out of every three to four

23

 

 

children before they reached their first birthday (Cornell 1996). Hence, before
industrialization, population grth was gradual in premodem Japan (Figure 2).
Famine and disease greatly contributed to this slow population growth, as did social
control - deliberate behavior to limit population growth, or consequences of
relatively long-term breastfeeding practice and migrant workers (dekasegi) (Cornell
1996; Hanley 1972; Hayami and Kurosu 2001).

The household, not the individual, was perceived as the basic social unit and legal
unit of the society (Hanley 1985; Saito 2000a). The title to rights in the village
community was attached not to the head of the ie as an individual, but to the ie. This
was because during the Tokugawa period, the peasant household was an economic
entity - the unit of economic production and consumption — and the family members
included kin as well as tenants and servants (Kumagai 1995). From the paddy field,
to the forest, and to dry land, these rights were attached to the ie rather than to any
individual family member (Cornell 1987). The increase of surplus productions owing
to a significant increase in agricultural productions during the Tokugawa period,
however, gradually enabled more people (e.g., more employees and their families,
and more relatives) to establish their own ie (Kasugai 2004). Consequently, the ie
began to shift its meaning from a social group based on cooperation in the ie
economy to a group based on kinship ties or the Confucian dyad of father and son
(Kasugai 2004).

Kumagai (1995) explained that among the lower strata, including peasants,

artisans, and merchants, and eta (35v) meaning “pariahs of society”, systematized

24

succession of the ie did not exist. Although all the people except eta had names
given to show their occupations, they were not family names. The headship of the
household was not restricted to the eldest son: each household selected a son of
superior character to guarantee the succession of the household (Kumagai 1995).
Among the upper strata, aristocrats, the shogunate, samurai warriors, and the
increasing numbers of independent farmers, the ie, gradually formed from the late
tenth century to the eleventh century, matured in the Tokugawa period into the
feudal family structure of the ie system. As the succession of the ie became more
important, systematizing the ie succession became a8 a promising way to pass on the
ie from one generation to another.

Under the ie system, one son, preferably the eldest son, remained in the parents’
household, while others had to leave. Family members were divided into those on the
stem family line and out of the line, and son-heirs enjoyed a higher social status than
other siblings (Saito 2000b). Because the marriage of the eldest son became the most
important and urgent matter for the succession of the ie, those who stayed in the ie
tended to marry earlier than those who left. Whether the other brothers stayed in the
household also depended on how large and extensive the ie estate was. To sustain the
material wellbeing of the collectivity and to adjust the size and structure of the ie to
meet the demands of family farming, parents sent children to work for someone else
such as merchants, kept in or removed other siblings from the household, took in an
employed farm laborer, or deliberately or unintentionally limited family size (Saito

2000b). As was common in pre-industrial Europe, one of the sons would marry and

25

take over the farm. In pre-industrial Europe, however, this process took place at the
retirement of the father (Davis and Blake 1956). In Japan, the old household head did
not need to retire when his heir married (Saito 2000b). Some elderly in the upper
strata held power and autonomy until or even after they retired from work to help the
ie succeed and to manage its members.

Although primogeniture was prevalent among the upper strata, the succession of
the ie itself was not an easy task. Children, especially boys, had a vested interest in
the continuation of the ie throughout time. Because of the lack of vaccination and
medical treatment in those times, however, infectious diseases often killed the eldest
son. Hence, demographically, it was necessary, especially among people in lower
strata, to keep the ie as a household based on economic cooperation and coresidence,
and not to restrict the position of heir to the eldest son. Because of the lack of
vaccination and medical treatment in those times, elderly parents were less likely to
rely on their children for a long time. According to Kito (2000), children needed to
take care of their parents for only about three years after their retirement.

After the beginning of the eighteenth century, writings about the morality of care,
concrete ways of caring for the elderly, and health began to increase. Hence, the
importance of care prevalent among the upper strata of society began to be socially
recognized (Yanagiya 2001). Ideologically, the Tokugawa shogunate emphasized the
Confucian ideology of filial piety. Several writings of Ekiken Kaibara, a Confucian,
emphasized the moral importance of children, especially the man as the head of the

ie, to care for the elderly. There are various kinds of such writings for men, and some

26

of them include concrete ways to take care of the elderly. There are also moral
writings for women, but they did not include such practical information about care
(Kasugai 2004). Some people in the upper social strata also hired servants to care for
the elderly, but these servants were not restricted to women. Some wealthy families
hired male servants to care for the elderly in the case of emergencies, including a fire
(Yanagiya 2001).

In addition, the Tokugawa shogunate also made an effort to strengthen and spread
this ideology to the people in the lower strata by providing such writings about filial
piety targeted at the populace. Owing to an emphasis on literacy for commoners as
well as the samurai, and the development of popular folk schools (tera koya:

$TE), it is estimated that close to 40 percent of the Japanese were literate before

1868, a very high figure by the world standard (Fuse' 1975). Among these writings,
more works again emphasized and valued the moral necessity of men, rather than
women, to take care of the elderly through coresidence (Kasugai 2004). According to

Ko‘g'roku (ififi) , the popular moral writing describing the importance of filial

care of the elderly, the father as head of the household sits at the head of the table,
often being replaced after his death by his wife, not his son, even among families in
the lower strata (Sugano 1999). This practice reflects the spirit of the Edo period
based on Confucianism.

The heavy duties of elder care, even when elders did not live as long as they do
today, are suggested by stories in Kogiroku. A man remained celibate to take care of

his parents. A woman’s parents, worried about her heavy duty of care, suggested her

27

that she divorce, but she refused. A man who took care of his mother divorced his
wife because she was not obedient to his parents (Sugano 1999; Yanagiya 2001).
Hence, Kogiroku not only highly valued women who were obedient and cared for
their husbands and his parents, but also highly praised men who financially,
emotionally, and physically took care of their parents.

Recently, Kyoto Prefectural Library and Archives (2005) reported that the village
of Kameoka allowed elder care leave for samurai. There was a letter written by a
firefighter in Kyoto in 1819, asking to take elder care leave to take care of his
grandmother. Hence, in the Tokugawa period, caregiving was not restricted to
women (Kyoto Prefectural Library and Archives 2005). There was no cultural or
universal ideal that daughters-in-law should be the main caregiver for the elderly.
Rather, care was often emphasized as part of men’s morality, a8 a responsibility
assumed by a head of household. As Kasugai (2004) pointed out, care was also
valued by the public, because it represented the Confucian precept of filial piety and
the spirit of the Tokugawa shogunate. It was only after the unification of the Meiji
period that the morality of care shifted from men to women, and from public to
private.

Emerging from 150 years of civil war, the Tokugawa period brought 150 years of
peace, which transformed the samurai class during the Tokugawa period. At the
beginning of the Tokugawa period, the samurai class was the real warrior class.
Without actual battles and separation from the land, however, the samurai class was

gradually transformed to a bureaucratic elite class of civil administrators living in

28

urban areas. Transformation of samurai into civilian administrators increased the
numbers of schools and literacy. Without actual battles, the number of the samurai
class also increased to two million people by the time of the Meiji restoration; the
number was larger than that of the English or European aristocracy (Tipton 2002).
Consequently, more ranks were created within the class and resulted in
bureaucratization of the samurai class. Bureaucratization also gradually transformed
the nature of loyalty and the lord-vassal relationship (oyabun and kobun) to
impersonal loyalty to the office. Accompanied by increasing costs of living,
maintaining the samurai in various rankings resulted in samurai impoverishment.
Hence, not from below but from within the ruling class, lower-class samurai who
were dissatisfied with the Tokugawa system led the struggle for the Meiji restoration
(Tipton 2002). Farmers suffered from an increasing burden of land tax as well. From
1824 to 1836, frequent bad weather led to famine, and the dissatisfaction of starving
peasants increased toward the rich who taxed and relied on food they cultivated
(Tipton 2002). Finally, the Tokugawa shogunate fell, and the Meiji Period began in

1868.

The Meiji Period to the Secon_d World War (1868 AD — 1939 A.D.I:

Modernization of the State and the Ie System
The new Meiji government achieved a centralization of political authority

unparalleled in history, and it pushed industrialization and economic growth to catch

up with and to build national strength to ward off American and European

29

 

 

 

imperialism (Fuse' 1975). Meiji leaders encouraged the adoption of Western social
customs and cultural styles to reach the goal of “enlightenment and civilization”

(bunmei kaika: 1535MB) (Tipton 2002). To become a modern state, Japan also

opened its markets to foreign trade and began to industrialize. These changes aided
economic and agricultural development. Cholera, however, frequently attacked the
Japanese people and killed thousands in 1886. One record indicated that the Japanese
government purchased rats from people to reduce cholera because they realized that
a flea parasite on the rats brought the cholera bacterium to Japan (Hayami 2001).
Due to crowded working conditions, diseases, such as tuberculosis, increased
with industrialization. In addition, endemic and infectious diseases were more
common than chronic degenerative diseases. In 1920, the leading causes of deaths
were pneumonia, bronchitis, gastroenteritis, and tuberculosis. Chronic degenerative
diseases such as heart disease, cancer, and stroke did not replace them until 1958.
Owing to gradually improving living standards, medical technology, and sanitary
conditions, however, the Japanese population gradually transformed from the stage
of high mortality and high fertility to the stage of low mortality and lowered fertility
during the Meiji era (Hayami 2001). The mortality rate fell faster than the fertility
rate, and so population growth accelerated and provided an expanding labor force
(Kito 2001). This course of industrialization was also a course for militarization and
nationalization in Japan. The concept of the ie became an unparalleled means for the

government to unify and strengthen its population.

30

Before 1898, there was no legal system governing people’s behaviors except for
the upper classes, including aristocrats and samurai (Asai and Kameoka 2005). The
ie was legally recognized in the Meiji Civil Code in 1898 and officially called the ie
system (ie seido). The ie system came to characterize both lower class as well as
upper class families in Japan (Kumagai 1995). The purpose of the Meiji Civil Code
was to recognize the family system, the pattern of primogeniture, as the basis for the
Meiji government to rule over people. Based on the Confucian teachings of filial
piety, the ie system formed the ideological ground for the Meiji government that put
the emperor as its head of the state and the state as a single family (Asai and
Kameoka 2005). Along with the Confiician ethos of the father and son relationship
being preeminent over the husband and wife relationship and the father and daughter
relationship (Tarnney 1998), the relationships between parents and children and
between husband and wife were like the relationship between superior and
subordinate, suppressing individuality to preserve the harmony of the family (Asai
and Kameoka 2005; Kumabara 2000).

The ie system effectively functioned for the government to unify and control the
thoughts of an ever-increasing Japanese population. To catch up with foreign nations
and to become a modern nation, the government socially constructed motherhood

(bosei: Eli) as woman’s morality. Prior to the Meiji period, there were no

widespread cultural expectations for women to show maternal affection in raising
children. For instance, although women in an upper-class family were considered

tools to produce a male successor, their infants were often given to a nurse to raise

31

(Kano 2000). Under the slogans “Fukoku Kyohei (Emmi) to enrich the country
and strengthen the military, and “Umeyo, fuyaseyo (Ed) rive Oh)”, to give birth
and increase the population, abortion was largely prohibited so as to increase the
quantity of children. To improve quality, however, abortion became legal under the
National Eugenics Law in 1940 to save the life of the mother and for a woman
suspected to be carrying a genetically defective child (Harrison 1999). To compete
with foreign nations as a modern state, the government privatized the role of women
by suppressing and educating them to be nurturing, wise mothers, and tactically used
the £2 system to become one collective under the emperor. Women became
socialized to be nurturing mothers who contributed to increasing the quantity and
improving the quality of the population (Harrison 1999). As a result, Japan kept its
TFR higher than 3.0 until the end of the baby boom period in the late 19408 (Figure
3), which contributed to its population growth (Figure 2).

As the population rapidly grew and industrialization proceeded, the government
required more men and money to support nationalism and industrialization. The
government tactically used the ie system to supply these resources by privatizing the
ie and reconstructing the morality of care. In 1872, the Meiji government replaced
the religious investigation registers or the village population registers (shumon

aratame-ché) with the family register system (koseki: 75“). In the former system,

all the members of a village were registered by household, whereas in the latter
system, the vital events were recorded on the basis of the stern family (J annetta and

Preston 1998). In other words, the ie became increasingly recognized more as a

32

“family” based on blood-ties and marriage rather than as a “household” based on
economic participation. The ie system provided the eldest son a legal right to inherit
family property, thereby placing considerable moral obligation on the eldest son and
his wife to take care of his elderly parents. Tied to the Confucian precepts that
stressed the responsibility of a wife to be obedient and take care of her family
(T amney 1998), the wife of the eldest son became the primary caregiver for any frail,
dependent, or sick family members by becoming unpaid housekeepers, nurses, and
companions (Harris and Long 1999; Long and Harris 2000; Traphagan 2003; Wu
2002). Women’s morality was tactically emphasized to supply an obedient
population for the emperor under the myth of motherhood (Kano 2000) and to repay

limitless obligations (on: E.) to the revered elderly under the tradition of Confucian

precepts. The ethics of care became increasingly recognized as women’s morality in
the private sphere, rather than men’s morality in the public sphere (Kasugai 2004).
Consequently, the life course became gendered, and aging and the family became
universally idealized. The dependent elderly were now expected to place themselves
in the care of their eldest son’s family and under the hands-on assistance of his wife
(the daughter-in-law).

The idealization and privatization of the ie enabled men to work in industries,
such as machinery and metal manufacturing, to support booming capitalism and
nationalism. According to the 1909 central government survey of factories
employing five or more operatives, two percent of women worked in machinery and

metal manufacturing industries. Importantly, however, capitalism and nationalism in

33

Japan were also achieved through hidden women’s work. In fact, no less than 85.2
percent of them, mostly young unmarried women, worked in the textile industry,
which was a leading sector in the early development of industrialization (1 880 A.D.—
1940 A.D.) in Japan (Okada 1998). In addition, along with booming
industrialization, a great number of families engaged in trade or professional services
came to hire female domestic servants. In 1930, there were 710,000 maids in Japan,
calculated as one in every seventeen households (Okada 1998). These female
domestic servants and female textile workers were mainly younger girls, the majority
of whom were rural in origin (Okada 1998). Hence, the modernization,
industrialization, and nationalism of Japan were supported by what Coontz (2000297)
called “women’s altruism”.

From the Tokugawa period to the beginning of the Meiji period, from the upper
strata to the lower strata, more elderly became “priceless” that their status became
sentimentalized with respect in the public community. A couple of decades after the
Meiji period started, textbooks described the elderly as socially weak and dependent
on care, while valuing those elderly who were active and independent (Kasugai
2004). On the surface, based on the Confucian precepts of filial piety, the elderly
were highly respected because they were resourceful; but in reality, they were often
identified as those who were unable to adapt to the modernization process (Kasugai
2004). In contrast with the Tokugawa shogunate that was based on the Confucian
spirit of filial piety, the Meiji period represents the time the government was eager to

modernize and to nationalize Japan. As the nation became capitalized, supporting

34

one’s wife and children was considered more profitable than supporting the elderly.
Hence, although demographically more elderly became culturally revered from the
Tokugawa to the Meiji period, their cultural status suffered with the privatization of
the moral responsibility for their care. It could be possible to state that the peak of
the social value of the dependent elderly was in the Tokugawa period when the

ethics of care was publicly valued and recognized.

Postwar Japan -— The Path to Becoming the Most Long-Lived Nation in the

M

The ie system was legally abolished after Japan lost the Second World War. The
United States took over the nation to democratize and Westernize it in a short period.
As a result, Japan has gone through dramatic social and demographic change since
the Second World War. For example, Japan experienced a steep fertility decline alter
the baby boom faded out in the early 19508 (Figure 3). In 1947, it was high
(TFR=4.54); however, ten years later, it was 2.11, the replacement level. This steep
fertility decline was related to the legal abolition of the gender-biased ie system and
its replacement by a new family system based on equality. It led Japanese women to
legally claim equal rights in inheritance, education, and employment. More and more
Japanese women gained greater opportunity to attain autonomy and freedom (Atoh,
Kandiah and Ivanov 2004; Kumagai 1984). Enrollment of women in senior high
school increased significantly in the 19608. By 1975, the enrollment rate rose to 90

percent, reaching, by 1985, 97 percent (Shirahase 2000). In addition, a new

35

government policy instituted in the mid-19708 to encourage enrollments in higher
education, especially for women, effectively supported the increase in women’s
enrollment (Retherford, Ogawa and Matsukura 2001). Between 1970 and 1975, the
rate of women’s enrollment almost doubled, reaching 26 percent of age-eligible
women in universities and 22.9 percent in junior colleges (Shirahase 2000).
Enrollment ratios increased rapidly at the junior college level for women and at
the university level for both men and women. Nevertheless, this postwar education
reform especially benefited women. The proportion of those who graduated from
junior colleges or universities climbed steeply for both sexes, and by 1998, the
proportion became greater for women than men (Retherford et al. 2001). According

to Japan ’s Education a; a Glance 2005 , the entry rate to higher education reached

 

74.5 percent in 2004, and it was 75.1 percent for women (Ministry of Education
Culture Sports and Science and Technology 2005). In addition, as the proportion of
women among four-year college graduates increased, more of them chose fields such
as the social sciences, natural sciences, or engineering rather than the arts and home
economics (Atoh et a1. 2004). Along with rising qualifications, labor shortages,
associated wage increases, and the fast growth of the service sector enabled women
to enter paid employment. Although some deny the direct influence of education on
marriage (Retherford et al. 2004; Ueno 1998), higher education for both men and
women has a salient influence on the postponement of marriage by strengthening
individual autonomy, raising educational qualifications, and increasing the labor-

force participation of women (Retherford et a1. 2001). In fact, economic gains from

36

education (wage gain per additional year of schooling) have been greater for women
than for men. Average monthly wage differences between junior high schoolers and
university graduates also have been higher for women than men (43 percent for men
and about 60 percent for women). This wage gap indicates that the economic returns
to education have been higher for women than men (Retherford et a1. 2001) and
implies that college diplomas for women became instruments for furthering their
careers by raising their wage potential (Atoh et al. 2004).

As both men and women gained equal educational opportunities, attitudes
towards gender roles and expectations also started to change for both men and
women. According to Harris and Long (1999), those who agreed with the statement
that men should work outside the home and women should remain at home
decreased from 36.6 percent to 22.3 percent for women, and from 51.7 percent to
32.9 percent for men, between 1987 and 1995. In addition, not only did women’s
time become more expensive due to higher educational attainment and longer female
participation in the labor market, but also rising economic costs for the education of
children became the main burden for women of reproductive age (Atoh et al. 2004).
It created an antinatalist pressure to below-replacement TFRs.

Although a larger share of Japanese women gained more education and
employment and married for romantic reasons, after they worked for a couple of
years, they found husbands who enabled them to climb up the social ladder, and then

withdraw from the formal labor force to become full-time homemakers (sengyo-

shufir: liifl). These women then took care of the domestic needs of their

37

husbands and children and his parents. Many companies congratulated women for
leaving jobs due to marriage (kotobuki-taisha: *igii), thereby pressuring single

women at work to follow this pattern. This pattern was reflected in the M-shaped
curve of female labor force participation in Japan (Figure 4). The lefi- hand peak of
the M indicates women entering the labor force after school, and the right- hand peak
of the M indicates their reentry into the labor force after raising a family (Ochiai
1996). Until Japan’s Equal Employment Opportunity Law went into effect in 1986,
most female workers were assigned to different career tracks than male workers.
Gender-segregated job tracks in large firms assumed that women would resign at
marriage or childbirth, and excluded women from promotional opportunities and
pressured them to quit their jobs on maniage or pregnancy (Lee and Hirata 2001).
Based on the 1991 National Survey of Occupational Mobility and Careers of
Women, Lee and Hirata (2001) found that a husband’s college education increased
the probability of his wife’s job separation. When men brought enough income to the
home, companies and families expected women to quit their job after marriage to
raise their family. This resulted in uniformity in fertility. Ochiai (1996:41) called this
fertility trend “the two-child revolution”. Although the majority of women during the
Meiji (1868-1912) and early Taisho period (1912-1926) had four or more children,
there was no uniformity in the number of children that couples had; but there was
increasing uniformity in the number of children that couples had among women born

in the early years of the Showa period (1926-1989).

38

As the economy experienced gradual but long stagnation, however, and the
divorce rate mounted, women increasingly realized the fragility of this modern
nuclear family. More of them postponed marriage to pursue higher education and
jobs to help them to climb up the social ladder, to enjoy their freedom by using their
time and money for leisure and hobbies, and to equip themselves to be self-
supporting if their marriage failed. Between 1975 and 1995, the mean age at
marriage increased from 24.5 to 27.7 years for women, and 27.6 to 30.7 years for
men (Retherford et a1. 2001). The office of Gender Equality in the Prime Minister’s
Office enacted the Basic Law for Gender Equality in 1999 to encourage companies,
schools, and communities to provide equal opportrmity for both men and women to
participate in all kinds of social activities and to share responsibilities. The
effectiveness of the Basic Law is, however, still questionable because there has been
a discrepancy between the law and an unchanging environment due to conservative
political elites who attempt to protect traditional family values that pressure women
to marry and have children.

By 2003, Japan’s TFR dropped to a record low of 1.29 (Figure 3). Sustained
fertility decline in Japan is now lower than the replacement level of 2.1, or what
demographers call the “the Second Demographic Transition” (van de Kaa 1987).
This indicates that postwar social reform, especially the postwar education and
family reform, restructured Japanese society to the point at which marriage has
become an individual choice over a traditional responsibility to preserve the ie. The

decline in arranged marriage, the increase in love marriage, the rapidly increasing

39

divorce rate, and the increasing prevalence of cohabitation and premarital sex
(Ochiai 1996; Retherford et a1. 2001) reflect the postmodernizing of Japanese society
and families that emphasizes individuals over collectives. Marriage and childbearing
became less appealing for women, especially for the highly educated, and increasing
economic independence contributed to a further postponement of marriage and to
permanent celibacy (Raymo 1998). These changes in patterns of marriage have made
it harder for eldest sons to guarantee hands-on helpers to their elderly parents.

At the same time, economic modernization has helped to raise Japan’s life
expectancy at birth (LEB) to the highest in the world: 80 years for males and 87.5 for
females (Figure 5). This achievement arose from an epidemiological transition—a
shift from infectious to chronic diseases (such as cancer and cardiovascular disease)
as the principle causes of death (Omran 1971). The rise in LEB in Japan from the
early 19208 to the mid-19608 was largely due to a reduction in the proportion of
deaths from infectious diseases (especially gastrointestinal disorders, tuberculosis,
pneumonia, and bronchitis) that were particularly fatal to infants and the elderly. A
main reason for the decline in fatalities from these diseases was the introduction of
antibiotics (Mertens 1994). As a result of this trend, chronic degenerative diseases
began to replace infectious diseases as causes of death (Martin 1989). The top three
causes of death in Japan are now (in the following order) malignant neoplasm
(cancer), cardiovascular (heart) disease, and cerebrovascular disease (stroke)

(Tanaka and Johnson 2006).

40

Such a dramatic demographic transition in Japan, caused by postwar
transformations in familial and medical institutions, led Japan to become the most
long-lived nation in the world in 2005, having the longest life expectancy at birth,
the largest portion of the population elderly (25 percent), and the smallest proportion
15 years and younger (13.6 percent) (Ota 2006). At the same time, more and more
women began to express their frustrations over being expected to focus most of their
energies on caregivng (Schoppa 2006). It brought the question of eldercare into
public discourse. Children, especially the eldest son and his family, are persistently
expected to repay the priceless obligations (on) they received from parents when
they were young through providing care in their old age as well as to continue the
name of the ie throughout generations.

These expectations have burdened women as primary caregivers for their
families. The limited supply of childcare and eldercare services, however, along with
tax and benefit rules subsidizing “dependent” housewives (sengya-shufir) through
providing them with pensions, tax credits, and other benefits, pushed women outside
of the formal labor market and pulled them into the ie. These shifts supported the
postwar economic expansion and growing capitalism of Japan (Schoppa 2006). Such
a system of capitalism supported by patriarchal ie has come under pressure as a
result of demographic and socioeconomic changes, such as the changing aspirations

of women.

41

Conclusion

To understand the social organization of intergenerational caregiving in Japan,
scholars must understand its historical transformation. Prior to the Meiji period, there
was regional and class diversity in family life. There was no universal cultural ideal
for women to take care of dependent elderly. As Japan moved from the Tokugawa to
the Meiji period, the decline in mortality enabled more sons to survive to adulthood
to continue the ie. It became more feasible for the national government to create
legal pressures relegating women to the home to care for children and the dependent
elderly in the privatized ie. Although the ethics of care moved from the public to the
private sphere, and from the contextual to the universal, socioeconomic
transformation enabled women to gain equal education and employment and enabled
the elderly to live longer. Generational differences in residence, occupation, and
education brought various conflicts and tensions in current Japanese society in
determining to whom the morality of care should belong.

“Who should take care of the ever-increasing elderly in Japan?” became the major
concern in the twenty-first-century Japan. Recent studies of aging and the family in
Japan involve intensive discussions of care for the frail and dependent elderly,
tossing the responsibility and burden between the public and the private spheres.
Prior to the Meiji period, care was valued in public, which shifted to the private
sphere to modernize, capitalize, and nationalize Japan within a short period. Rapidly

decreasing fertility and ever-increasing numbers of elderly in postwar Japan

42

provoked intensive discussions of care, which brought back the issues of care to
public awareness.

Throughout its history, care meant the caregiver’s process and obligation of
looking after someone weak and dependent, and care receivers were perceived as
passive recipients. Today, owing to longevity and reduced disability among the
elderly after the Second World War, more elderly have challenged such a pessimistic
and passive view of care and they are coming to be seen as proactively choosing
their opportunities to care and to be cared for by important people in their later life
course, and the study of social integration in Japan becomes important to understand
and promote healthy aging in Japan. It is within this historical and cultural context
that Japan now faces the question of how to promote a harmonious integration (wa)
and mutual dependency needs (amae) of its senior and junior generations in later life

in the twenty-first century.

43

CHAPTER 3
THEORETICAL FRAMEWORK
Aging and Family in Japan in the Twenty-First Century:

Social Recognition of the Importance of Social Integration on Health

Introduction

Today, more elderly have challenged negative stereotypes toward them, and they
are seen increasingly by scholars as proactively choosing their opportunities to care
and be cared for by people important to them during their later life course. Reflecting
such social recognition of the importance of social integration on healthy aging,
research on social integration and health has emerged as an urgent and important
issue of both public and scientific concern. This chapter first describes why studying
social integration in later life became important in studying aging and the family.
Second, theoretical frameworks explaining the relationship between social
integration and health and empirical studies are reviewed. Finally, this chapter ends
with three research questions based on the literature review of theoretical

frameworks and previous studies of social integration and health.

Why Social Integration is Important in Studying Afl'ng?

Japan became the nation with the greatest longevity in the world in 2005, having
the longest life expectancy at birth (82 years in 2006), the largest portion of the

elderly (25 percent), and the smallest proportion of the population 15 years and

44

younger (13.6 percent) (Ota 2006). Kaneda and Raymo (2003) reported that over the
past few decades, the main source of increasing life expectancy shifted from
reductions in mortality due to senility to cerebrovascular disease, and then to heart
disease. Particularly, declines in mortality from cerebrovascular disease over the last
three decades suggests socioeconomic and behavioral factors, including better
housing and a more balanced diet, played an important role in achieving an increase
in postwar life expectancy; improvements in medical treatment also contributed to
the increase (Kaneda and Raymo 2003). Japan also experienced a steep fertility
decline after the baby boom faded out in the early 19508. In 1957, it was high. The
Total Fertility Rate (TF R), the estimated average ntunber of children born to a
woman in a lifetime, based on the age-specific birth rates observed in a given year,
was 4.54. However, ten years later, the TFR was 2.11, the replacement level. Ever-
increasing numbers of elderly along with rapidly decreasing fertility in postwar
Japan provoked intensive discussions about care, and brought those issues into
public awareness. One of the most frequently asked questions about care brought to
the public is, “Who is to take care of the ever-increasing numbers of elderly?”

The cultural superstructure defining the proper organization of eldercare has
changed over the course of history, because of new interactions among demographic,
economic, and political realities (see Chapter 2). In the Tokugawa period, various
writings emphasized the moral importance of a man to care for the frail elderly as an
important practice of filial piety. Such morality of care shifted to the private sphere

and women’s arena during the Meiji period in order to modernize, industrialize, and

45

nationalize Japan within a short period. Along with the legalization of the
primogeniture, women, especially the wives of eldest sons, became the culturally
idealized main caregivers of the dependent and frail elderly.

The major weakness of the previous literature on aging and family in Japan is that
scholars often start with describing this ie system as the benchmark, assuming that
the cultural underpinnings of the ie define the care of ‘frail elderly’ as ‘women’s
work,’ undertaken to preserve ‘women’s morality.’ Many scholars discussed
women’s burden and responsibilities to care for the frail and dependent elderly in
patriarchal Japanese society from the standpoint of caregivers. This resulted in an
intensive focus on the discussion of eldercare as a persistent burden on women,
while explicitly or implicitly categorizing care receivers as passive recipients. Such a
“pessimistic” view of the elderly has been enhanced by the media and some
specialists stressing the seriousness of the problems of old age (Koyano 1997). In
such a context, since care has been perceived as a caregiver’s responsibility and
obligation for looking after someone weak and dependent, the elderly have too often
been associated with burdens and problems, stereotyped as sickly, frail, and
dependent.

The idea that older persons cannot maintain active participation in society and the
economy appears to be more myth than reality. Increasing longevity also challenged
the "elderly myth" and various studies showed reduced disability among the elderly.
US. studies reported that the decline in morbidity, lower incidence of disability,

higher rate of recovery, and significant improvement in ability to work at older ages

46

have accompanied the increase in longevity at older ages (Crimmins, Reynolds and
Saito 1999; Crimmins, Saito and Reynolds 1997; Manton, Stallard and Corder 1995).
Japan also experienced similar gains experienced in the US, with annual declines in
disability and functional limitations ranging between 1.5 percent and 3.5 percent
(Schoeni et al. 2005). Japan also experienced improvements in mortality that began
largely in the 19708 (Schoeni et a1. 2005). According to Schoeni et a1. (2005), Japan
experienced substantial decline in disability among the elderly over the past decade,
reducing the number disabled being by about 1 million or 14 percent relative to what
the numbers would have been if the disability prevalence remained the same; these
reductions were experienced across economic and socio-demographic groups. These
findings weakened the idea that Japanese elderly are dependent on women's morality
located in the private sphere.

Owing to their longevity and reduced disability among the elderly, more Japanese
elderly have challenged negative stereotypes. The elderly are seen increasingly by
scholars as proactively choosing their opportunities to care and be cared for by
important people in their later life course. Although the government remains
criticized for heavily relying on families for filial care for the elderly, there have
been increases in the quantity and quality of private and public care facilities, various
social activities, and volunteering programs for the elderly provided by local
governments and non-profit organizations. These facilities, programs, and activities
aim to provide opportunities for the elderly to remain in the circle of harmonious

integration (wa), where people can care for each other beyond the family. In

47

addition, since Japan has universal health coverage with set benefits including
physician and hospital services, dental care, and prescription drugs, local programs
to improve health and functioning, to postpone disability, and to promote recovery
from morbidity have become, and will continue to be, crucial factors to stem the rise
in government budgets (Schoeni et al. 2005).

Several reforrnations of the social security system have also changed the ability
and the necessity of the elderly to care for themselves. Under the ie system, caring
for the dependent elderly by family members was legitimized under the then-barely
developed social security system (Thang 2000). Between 1965 and 1994, social
security benefits, which encompass old-age pension schemes and medical plans,
increased fi'om six to sixteen percent of the national budget. Between 1981 and 1996,
of elderly Japanese aged 60 and above, the proportion that received public pension
benefits increased from 65 percent to 84 percent, and the proportion which reported
their primary source of income to be their pension increased from 35 percent to 57
percent (Ogawa and Retherford 1997). However, to meet their budgets with ever-
increasing numbers of elderly, this pension scheme was revised in the 1999 Pension
Reform Act, which includes a 5 percent reduction in pension benefits, and an
increase in the eligibility age from 60 to 65 years old (Oshio and Oishi 2004).
Unlike the United States, Japan is reluctant or slow to rely on large-scale
immigration to fill gaps in its labor force (Moffett 2005). Thus, for the government,

enticing the elderly to keep active and to work longer before receiving retirement

48

benefits is considered a strategy to solve the labor shortage and to alleviate financial
pressure to support the ever-increasing numbers of elderly.

In fact, in 2004, Japan passed a law that requires companies by 2013 to raise their
retirement age to 65 from 60 or rehire their retiring workers (Moffett 2005). Maeda
and Ishikawa (2000) reported that 78.9 percent of older men aged between 60 and 64
were working and 41.9 percent at age of 65 and above, which is far higher than for
the United States, Denmark and Germany. According to the economist Atsushi
Seike, the country’s lower birth rate indicates the number of Japanese in their 208
will decline by 3.2 million over the next decade; however, extending everyone’s
working life until 65 will keep two million or more in the work force (Moffett 2005).
Oshio and Oishi (2004) claimed that pension reform is expected to increase the
average retirement age by about one year for men, which implies that more elderly
will continue to stay in the labor force after official retirement at age of 60 or 65. In
Japan, an elderly person's labor force participation is higher in rural areas than in
urban areas. Forty-one percent of rural males were in labor force in 1999 as opposed
to 32 percent of urban males. Similarly rural women showed higher labor force rates
than urban women (Ogawa 2004). Since rural elderly workers tend to keep working
until their health condition deteriorates, in contrast to urban elderly workers who are
more likely to be in the highly organized labor market and thus expelled at an earlier
age, the effect of pension reform on the elderly is expected to be greater for urban
residents. These policy initiatives, to remain solvent, will benefit from public health

policies aimed at extending the healthy life expectancy of elder adults in Japan.

49

Finally, although Japan experienced declines in disability similar to the U.S.,
Japanese society is often in contrast to, or lags behind, Western nations in terms of
balanced gender roles in the labor market. Nevertheless, more than ten years of
continuous recession and job insecurity have created greater pressures on women to
supplement their husband's earnings in a significant way (Takeda et a1. 2006). The
Equal Employment Opportunity Law promulgated in 1986 and the Revised Equal
Employment Opportunity Law in 1999 also emphasized and promoted equal
opportunities for men and women (Mori et a1. 2002). In such a context,
postponement of marriage and low fertility partly reflect the revolt of women against
cultural expectations that they be caregivers at home (Takeda et a1. 2006). Therefore,
not only has the increasing number of elderly Japanese challenged the myth that they
are incapacitated and needy, but also the increasing number of women has
challenged the myth of women's morality located in the privatized ie.

Reflecting these social and demographic changes and needs, increasing numbers
of elderly have started to show their preference to remain independent in Japan.
After the new Civil Code in 1946 replaced the ie system with the conjugal family
system, industrialization, democratization, and urbanization increased the mobility of
younger generations, which contributed to the nuclearization of the family (Thang
2000). Mass media also exposed Japan to Western ideas, including ways of looking
at caregiving (Hashizume 2000). Although the proportion of the elderly living in
three-generational households is still higher than in Western countries, “szipu no

samenai kyori” (a distance which does not allow the soup to get cold) became a

50

popularized metaphor as an ideal form of family-based support for the elderly
(Kweon 1997). Social commentators and mass media suggested this ideal distance to
reduce conflicting values between traditional family-based care for the elderly and
postwar socio-demographic changes. Such popularized metaphor introduced by the
media began to weaken the unquestioned romanticized image of Daikazoku, meaning
a family with large number of members, in which the elderly are taken care by their
offspring.

This “ideal distance” emphasizes familial love based on traditional
intergenerational relationships within the family; however, it does not necessarily
require intergenerational coresidence. Through fieldwork in an urban city in Japan,
Kweon (1997) found familial love based on traditional intergenerational
relationships within the family without coresidence. An elderly woman enjoyed her
independence by joining in various social activities for the elderly, while enjoying
her time by occasionally sharing family meals and exchanges of information. Kweon
(1997: 372) interpreted this distance of “togetherness and separation” as a result of
both generations trying to maintain a balance of independence and closeness. Thus,
“ideal distance”, in turn, challenges the stereotype of elderly Japanese as dependent,
frail, and demanding of constant hands-on assistance. In fact, the number of older
people living independently but close to their children in the same community
appears to have increased in recent years (Maeda and Ishikawa 2000).

In Japan, only eight percent of the elderly aged 60 and over live alone compared

to about 40 percent in the United States and Germany; it is one of the significant

51

differences between the lives of the elderly in Western nations and Japan (Maeda and
Ishikawa 2000). However, if this ideal distance is taken into account, the difference
between Japan and Western nations becomes much smaller. Indeed, in many
Western nations, about half of older people are living separately from their children,
but they live geographically close to them. Maeda and Ishikawa (2000) report that a
higher percentage of elderly Japanese lives separately from their children in large
metropolitan areas than in rural Japan, partly a result of the limited size of available
houses in urban areas. They also claim that living arrangements of the Japanese
elderly will change gradually and significantly in the future since the proportion of
middle-aged people who prefer to live separately is greater than the older generation
(Maeda and Ishikawa 2000).

Fieldwork in another urban city of Japan by Thang (2000) showed that
increasingly local communities are creating space for the old and young to interact
through community-based programs, which have prevented social isolation of the
elderly (Maeda and Ishikawa 2000). Such communities enabled the elderly to gain
emotional support beyond intergenerational coresidence and without being solely
dependent on their family members. Based on fieldwork in a rural village in Japan,
Traphagan (2004) found that older people were involved in activities at the
community centers such as the tea ceremony and “gate ball” (similar to croquet) in
order to avoid the onset of senility. He claimed that, to be a good elderly person in
Japan, one needs to be a “socially engaged individual who is involved in activities

that incorporate social interaction; failure to maintain activity and social involvement

52

invites loss of well-being” (Traphagan, 2004, p. 9). As Klinenberg (2002) claimed,
the recognized source of longevity has broadened beyond just the maintenance of
physical and mental vitality to include the social integration of older people with
family, neighbors, and other fiiends. In such a context, social integration can be
symbolized by, but should not be limited to, physical coresidence with younger
generations of relatives.

Reflecting such social recognition of the importance of social integration on
healthy aging, research on social integration and health has emerged as an urgent and
important issue of scientific as well as public concern for promoting healthy aging in,

Japan as well as in other nations of the world.

Theoretical Framework: Understanding Social Integr_ation - Multiple Role

Theories

The classical sociologist who initiated the theoretical and empirical study of
social integration and disintegration was Emile Durkheim (Berkman et al. 2000;
Moen, Dempster-McClain and Williams 1989; Pillemer and Glasgow 2000). His
major aim was to explain how individual pathology was a function of social
dynamics, and his work showed how social integration can influence mortality
(Berkman et al. 2000). In his famous theoretical and empirical study, Suicide, he
found that suicide was related to an individual’s integration in society, i.e., in
institutions such as family and religion (Pillemer and Glasgow 2000). In his view,

social integration was a strategy against modernizing forces, and he tied modem

53

urban life to increasing alienation and breakdown of social cohesion (Kushner and
Sterk 2005). His study showed that suicide rates were higher for single, divorced,
and widowed people than for married people, and suicide rates were higher for men
without a wife and children than for men with a wife and children (Durkheim 1951).
Durkheim also argued that the low rates of female suicide were reflective of men’s
subsumption in traditional gender roles (Kushner and Sterk 2005). Based on his
gendered assumption that women were less involved in public life than men and thus
less likely to obtain either the benefits or the costs of that involvement, women were
more immune to suicide than men (Durkheim 1951; Kushner and Sterk 2005; Moen
et al. 1989).

After Durkheim, there have been many theoretical and empirical developments to
explain the positive relationships between social integration and health in later life,
and the most traditional way of theoretically and empirically explaining this
relationship is to take role involvement as an indicator of social integration
(Wethington et al. 2000). To understand how roles influence health outcomes, two
different explanations have been offered: the role strain perspective and the role
enhancement perspective (Barnett and Hyde 2001; Barnett and Marshall 2001;
Kikuzawa 2006; Moen et al. 1989; Wethington et al. 2000).

The role strain perspective argues that multiple roles could entail obligations that
produce overload and strain, which in turn could lead to negative health outcomes,
including psychological distress (Barnett, Marshall and Pleck 1992; Kandel, Davies

and Raveis 1985; Voydanoff and Donnelly 1999). This perspective is based on

54

structural-functionalism and focuses on the analysis of social structure. In this
perspective, social structures are viewed as made up of roles, which are the
behavioral enactment of a position in a social system coupled with behavioral
expectations and obligations (Goode 1960; Kikuzawa 2000). The role strain
perspective assumes that individuals are actors playing assigned roles embedded in a
position, which is predetermined by an institution (Goode 1960; Kikuzawa 2000).
Merton (1968) used the concept, status set, to describe the complex positions
assigned to individuals in various social systems. Multiple roles is a concept
isomorphic with status set (Moen et al. 1989), defined as behavior that fulfills
multiple positions under responsibility, expectation, and rights. The role strain
perspective maintains that individuals engaged in complex positions are more likely
to experience distracting and conflicting role obligations because they cannot fully
play all the roles following the specific scripts written for each role (Goode 1960).
By contrast, the role enhancement perspective states that the greater the number
of roles, the greater the level of social integration and connectedness (Moen et al.
1989). This perspective suggests that engaging in multiple roles, rather than fewer
roles, leads to higher levels of physical and emotional health, because the
accumulation of social identities or roles benefits individuals (Wethington et al.
2000). This perspective was developed from symbolic interactionists’ critique of
functionalist assumptions that individuals play roles in scripts written by the society.
This perspective assumes an individual is not only as a role-player but also a willful

actor who engages in role bargaining through interacting with others (Kikuzawa

55

2000). The role enhancement perspective emphasizes that holding multiple roles
does not inevitably cause role conflict or role overload, but it can benefit the health
of individuals by enhancing individual resources, social connections, prestige,
emotional gratification, and social identity (Barnett and Marshall 2001; Chrouser and
Ryff 2006). This perspective also emphasizes that an individual who occupies
diverse roles can fall back on another relationship if she or he encounters difficulties

in a particular role (Chrouser and Ryff 2006; Kikuzawa 2000).

Social Int tion and Health of the Elder : Role Enhancement or Role Strain
Previous studies supporting the role strain perspective targeted women in the
childbearing ages or middle-aged women, not the elderly. These studies discussed
how juggling their marriage, childbearing, and job caused poor mental health among
female workers who are in the childbearing ages or middle-aged (Chandola et al.
2004; Matsui, Ohsawa and Onglatco 1995; Mori et al. 2002). In addition, earlier
research on family roles on the well-being of the elderly were often restricted to an
examination of having a spouse, parent, and job. In particular, studies focusing on
why married individuals live longer and healthier lives than singles became very
popular in past decades (Gove 1973; Horvvitz, White and Howell-White 1996; Hu
and Goldman 1990; Rodgers, Hummer and Nam 2000; Ross, Mirowsky and
Goldsteen 1990; Zhang and Hayward 2006). These scholars argued that marriage
increases social support and reduces unhealthful and risky behaviors such as failure

to maintain an orderly life style, and their empirical studies showed that marriage

56

reduces causes of mortality such as tuberculosis, diabetes, and other infectious
diseases, and promotes a healthful lifestyle (Horwitz et al. 1996; Rodgers et al.
2000). By contrast, the divorced and widowed have higher death rates from heart
disease, stroke, pneumonia, many kinds of cancer, automobile accidents, homicide,
suicide, and cirrhosis of the liver, which all are leading causes of death in the United
States (Ross et al. 1990). A recent study by Zhang and Hayward (2006) also found
the significant and positive influence of marriage on the physical well-being of a
later life in the United States.

In Japan, Hu and Goldman (1990) studied the relationship between marital status
and mortality and pointed out that a similar relationship can be seen not only in the
U.S., but also in 16 industrialized nations that include one East Asian country, Japan.
They found that the high mortality of Japanese singles is prominent compared to the
mortality of singles in other countries. In the mid-19008, the life expectancy of 20-
year-old Japanese men or women who married would be as much as 15 years higher
than that of 20-year old Japanese men and women who remained unmarried
(Goldman, Takahashi and Hu 1995). In 1940, single women in Japan had a life
expectancy between 17 and 22 years lower than the ages for single women in the
United States, Finland, and France, and 17 years lower than the age for married
Japanese women (Goldman 1993). In Japanese society, especially for the elderly,
marriage is an “essential stage” in the transition to adulthood, and singles may face
deprivations due to the general rejection of the single and divorced status in Japanese

society (Goldman, 1993; Goldman, Takahashi, and Hu, 1995; Dixon, 1978;

57

Kumagai, 1984; Harnabata, 1990). The reason why married Japanese are healthier
than unmarried Japanese could be also explained by selection, because an excess in
mortality of singles could screen out potential spouses with mental or physical
problems such as obesity, hypertension, and nicotine addiction (Goldman et al.
1995). However, these studies of marital status on the health of the elderly did not
consider the effect of multiple roles on the well-being, as well as the effects of other
family roles and roles beyond the family such as participation in social groups.

Recent studies of multiple roles on the well-being of the elderly emphasize the
importance of social integration beyond the family, reflecting the criticism of the
negative stereotype of the family as the major and the only source of social
integration for the elderly. In examining the relationship between social integration
and the well-being of the elderly, various scholars found evidence that the role
enhancement perspective is more applicable to older adults than the role strain
perspective (Adelmann 1994a, 1994b; Chrouser and Ryff 2006). They all claimed,
based on their research with the elderly, that role accumulation not only enhances
individual resources and identity but also deters social isolation and provides choices
to those disengaged from particular responsibility (Adelmann 1994a; Choi et al.
2007; Chrouser and Ryff 2006; Moen et al. 1989; Moen, Dempster-McClain and
Williams 1992; Rozario, Morrow-Howell and Hinterlong 2004).

For instance, Adelrnan (1994) found that occupying more rather than fewer roles
leads to better subjective health, fewer chronic conditions, and fewer health

disabilities for both elderly men and women in the United States. A recent study by

58

Barnett and Hyde (2001) found that more roles enhance the level of subjective health
and lower levels of mental and physical health problems. Moen et al. (1989, 1992)
found that holding multiple roles could be a mediating factor in the negative
relationship between health and socioeconomic status, by maintaining or expanding
networks of family members and friends. In particular, they found participation in
volunteer activities as well as multiple roles showed a significant effect on health,
although their sample was restricted to women (Moen et al. 1992). However,
because most of the previous studies are based on Western contexts, little is known
about these relationships in non-Westem contexts (Comman et al. 2003).

In East Asia, several studies support the applicability of the role enhancement
perspective to understanding of the well-being of the elderly (Beckett et al. 2002;
Comman et al. 2003; Krause et al. 1999; Sugisawa, Liang and Liu 1994; Wu and
Rudkin 2000). For instance, Beckett et al. (2002) found that poor health status is
associated with low socioeconomic status, not having any living children, low
participation in social activities, and limited networks of friends in Taiwan. Comman
et al. (2003) found that negative perceptions of social support are significantly
related to lower levels of depression in Taiwan. Wu and Rudkin (2000) found that
social integration plays an important role in reducing the negative effects of low
socioeconomic status on health in Malaysia. Their study found that low
socioeconomic status was associated with poor health for all three ethnic groups
(Malay, Chinese, and Indian). However, they also found that this negative effect

tends to be stronger for older people with less frequent contact with their adult

59

children than for older people with more frequent contact with their adult children
(Wu and Rudkin 2000).

In Japan, Sugisawa et al. (1994) found that social participation, operationalized in
terms of organizational attendance, showed a significant direct effect on mortality.
Those with any degree of social participation had a risk of dying which was 68
percent of that for those who lacked social participation (Sugisawa et al. 1994).
Findings of Krause et al. (1999) indicate that the elderly who provide assistance to
others tend to rate their health more favorably than the elderly who are less involved
in assisting others. These studies show the importance of social integration on the
well-being of the elderly is not restricted to the family. However, they did not
consider the effect of multiple roles on the health of the elderly.

Kikuzawa (2006) found that occupying more roles was associated with lower
levels of depression for both American and Japanese men and women, and the
benefits of each additional role were greater in the US. than in Japan. She concluded
that this difference reflects the fact that “concentration of support resources in a
particular social group (e.g., the family) is more likely in societies that emphasize
collectivism, collectivism facilitates intimate interactions among group members
while discouraging the members from outside the group” (Kikuzawa 2006:73). Her
research refined the study of multiple roles on health by emphasizing the importance
of cultural contexts, and it is one of few studies that focuses on the relationship

between multiple roles and health in Japan. However, her study includes several

60

limitations in scope, many of which apply to the criticism of studies looking at the
effects of multiple roles on the well-being of the elderly.

First, Kikuzawa's study is based on national data from 1987, when women began
to remain in their jobs even after marrying and having children, a result of the
Japanese Equal Employment Law of 1985 (Matsui et al. 1995). At the same time,
Japan experienced economic growth in the 19708 and 19808, which was supported
by a gendered division of labor assigning women to be full-time housewives to
support men. In other words, her study was carried out prior to decades of a long
economic recession that steadily undermined traditional gender norms, even
traditional living arrangements (Takeda et al. 2006). Her data, collected 20 years
ago, is greatly limited in capturing the diversity of aging and the family in
contemporary Japanese society.

Second, because her study focused on a comparison between the US. and Japan,
rural and urban differences in social integration and health of Japan were ignored,
although spatial inequality is considered an important factor influencing social
integration and aging in Japan. Without considering spatial differences,
characterizing Japan as a nation marked by collectivism made her analyses
ambiguous.

Third, her study lacks measures that look at the quality of relationships such as
satisfaction with the amount of contact with their children. Fourth, her study argued
that accumulation of roles make the elderly more valuable and guarantee status

security; but these studies did not consider role density - whether or not having more

61

children and grandchildren matters. Finally, Kikuzawa's study did not involve an
analysis of time, i.e., whether or not (1) duration in roles such as marriage and
employment matters, and (2) whether or not engaging in multiple roles contributes to
better maintenance of physical and psychological health.

Finally and especially, an inability to incorporate the process of aging embedded
in a broader context is considered by scholars a major weakness of her study, as well
as previous studies looking at the effects of multiple roles on the well-being of the
elderly. Such criticism was repeatedly pointed out as the importance of research on
social integration and health increased in various fields, including public health,
medicine, and psychology (Wethington et al. 2000). In such a context, the life course
framework has emerged as an incorporating framework to advance research on social

integration and health in later life.

Incogporating the Life Course Framework into the Study of Social Intemtion
and Health

The life course framework is an approach to understand aging that emphasizes the
prior life experiences in determining later life outcomes (Elder 1985; Quandagno
2005). This framework provides a important lens to view successful aging because it
investigates pathways and how circumstances in early adulthood may affect social
integration and health in the later life (Moen et al. 1992).

In studies of the life course, the concept of trajectory and transition are central

themes (Elder 1985). Life trajectory can be described as linking states over

62

successive years such as a state of health. Each trajectory is marked by a sequence of
life transitions such as changes in the state of health that are better or worse (Elder
1985). In examining life trajectory of health and transition to a better or worse state
of health, a life course framework can go beyond the primary issue of multiple roles
and health (Moen et al. 1992). The life course framework suggests that the
connection between social integration and health can be viewed as a dynamic,
interactive, and possibly cumulative process, in which multiple or individual roles
previously or continuously occupied play out in diversity and density of social
integration over the life course, thereby sustaining a state of well-being (Moen et al.
1992). For instance, Moen et al. (1992) found that American women occupying more
roles in 1956 were in better health in 1986, controlling for background variables such
as age and previous health, than were women occupying fewer roles. On role in
particular, they found that being a member of a club organization significantly and
positively influenced health. They concluded that both social integration (in the form
of multiple roles) and health in later years reflect choices made and experiences
undergone throughout adulthood (Moen et al. 1992).

The life course theory assumes that individuals are agents, planning their own life
courses, but they build their own life courses as they relate to specific socio-
historical opportunities and constraints (Walker, Allen and Connidis 2005). At the
same time, the life course framework emphasizes the importance of the effects of
broader contexts such as gendered context, historical context, and geographical

context because individuals live their lives in such contexts, which produce

63

differences in roles, relationships, and resources in later life course (Elder 1985;
Walker et al. 2005).
o How does Gender Affect the Conditions between Social Integration and

Health?

Moen (2001) claims that men’s and women’s roles and resources are socially
constructed because of the different life pathways that exist in society. In her view,
socially constructed life pathways benefit men more than women since the expected
life course - education, employment, and retirement — more closely characterizes the
experience of men than women (Moen, 2001). Women who seek equality need to
accommodate these life course transitions while performing housework and care
work “traditionally” allocated to them (Moen 2001). Krause et al. (1998) found that
elderly women tend to report more problems with fitness items, to have more
symptoms of depression and to rate their health less favorably than elderly men do in
Japan. Supporting Durkheim’s discussion of women’s lower likelihood than men of
involvement in collective activities, Krause et al. (1998) argued that the subordinate
status occupied by women in public roles might generate stress due to their
disadvantaged status, and thereby exert an adverse effect on the health of women.
However, I argue, since Japan has the most long-lived women in the world
(Hashizume 2000; Takeda et al. 2006), socially constructed roles and resources in
the later life course may rather benefit women more than men in the later life course.

In looking at social relations in later life, scholars have argued that women tend to

have more social relationships than men, and that women's relationships tend to be

64

more intimate and involve a greater number of exchanges than men's (Antonucci and
Akiyarna 1995). In fact, women tend to receive more social support from the
government and their relatives than men do in Japan. Compared to widowed or
divorced men, widowed or divorced women can receive more financial protection
fiom the government through widows’ pensions in Japan (Ikeda et al. 2007). Such
socially and culturally constructed roles and resources in later life influence
mortality. Ikeda et al. (2007) found that the risk of mortality from cardiovascular
disease, respiratory disease, external causes, and all causes were two to three times
higher for never-married men than for married men. They also reported smaller but
significantly higher mortality risks from all causes for never-married women: higher
risk of mortality from cardiovascular disease, external causes, and all causes
associated with divorce and widowhood for men, but not for women (Ikeda et al.
2007). Hence, women may be better protected than men from the processes such as
widowhood in their later life course when it is placed in gendered context.
0 Does Age Influence the Relationship between Social Integration and Health?
Historical context influences people in various stages of life by affecting the
availability of resources from family members and their modes of assistance
(Easterlin 1995; Hareven 1994, 1995). Easterlin (1995) argued that parents of baby-
boomers will benefit more than parents of post baby-boomers, owing to the high
fertility of the postwar generations during the baby boom period. In other words,
parents of baby-boomers are more likely to have more children to provide social

support in later life than parents of post baby-boomers (Pillemer and Glasgow 2000).

65

In Japan, the older elderly are expected to have more siblings, children, and
grandchildren than the younger or future elderly, who are more likely to be
influenced by a shrinking family size due to declining fertility in postwar Japan. The
role enhancement perspective claims that role accumulation enhances one's self-
conception and self-esteem (Kikuzawa 2006). Thus, the life course placed in
historical context may benefit the older elderly more strongly than the younger
elderly in Japan since not only are the former more likely to have grown up with
more siblings and to live with children or grandchildren but also they are more likely
to have more of these descendants, who enhance resources, emotional support, social
connections, and social identity.

0 Does Rurality Ameliorate or Mitigate the Risks of Poor Health?

The life course approach also recognizes the importance of geographical context
because residential environment also shapes the context of individual’s lives
(Glasgow 2000, 2004; Thoits 1991, 1992; Wickrama et al. 1995). Previous studies in
the United States showed that rural men and women are more likely to subscribe to
traditional values and norms than urban men and women (Wickrama et al. 1995).
Accordingly, while social bonds may be tighter among ruralites than urbanites, the
greater poverty and fewer medical resources in non-metro U.S. counties, compared
to metro counties, may overrule the greater health and lower mortality that should
result fiom stronger social bonds (Hummer et al. 2004). According to Hummer et al.
(2004), individuals in non-metro areas had 80 percent higher odds of non-coverage

by any kind of health insurance and about 30 percent higher odds of Medicaid

66

coverage than those in metropolitan areas. Hummer et al. (2004) also reported that,
although individuals in non-metro areas are more likely to be married and less likely
to be never-married than those in metro areas, a higher level of poverty and lower
level of education are prevalent in non—metro areas (Hummer et al. 2004).

As such, any analysis contrasting mortality rates must control for the economic
disparities between non-metro and metro residents. For example, a study conducted
by McLaughlin et al. (2001) found lower rates of mortality in non-metro U.S.
counties than metro counties, despite substantial disadvantages associated with non-
metro residence. McLaughlin et al. (2001) claimed that the greater homogeneity and
the smaller differences in services and facilities in rural than urban areas reduce
feelings of deprivation and lead to a smaller effect of income inequality on health in
non-metro counties (McLaughlin, Stokes and Nonoyarna 2001). These factors may
symbolize and facilitate stronger social cohesion in non-metro America that would
produce the lower (than metro) mortality rates observed by these researchers over
several years using the 1990 US. census.

The associations among rurality, social integration, and mortality in the US. need
updating (Glasgow 2004; Johnson 2004). For example, Glasgow (2004) argued that
the scarcity and simplicity of social organizations in rural America offers fewer
opportunities for formal social participation in clubs and voluntarism, and fewer
public transportation services to travel to meetings than do those in urban America.
A study conducted by Johnson (2004) found no non-metro/metro spatial pattern of

the prevalence of disabling chronic diseases and disabilities. Obviously, whether

67

rural America offers a haven from medical risk through stronger social bonds
remains an unanswered question. This question is hard to answer because of the
substantial migration across nonmetro—metro boundaries by Americans over their life
course.

In Japan, rural areas symbolically represent a place which embodies “traditional”
Japan under the ie system. The rural elderly are more likely than the urban elderly to
coreside with an eldest child (particularly, the eldest son) and his or her spouse and
their children (Traphagan 2004). In addition, the rural elderly tend to work until their
health condition deteriorates in contrast to the urban elderly, who are more likely to
be employed in the highly organized labor market (Ogawa 2004) and thus
encouraged out of it before their health condition deteriorates. Fukuda et al. (2005)
found that the positive association among socioeconomic inequality and morbidity as
well as mortality of the elderly was buffered by mutual dependency (amae) and
harmonious integration (wa), especially in rural areas. And these findings are
consistent with these of McLaughlin et al. (2001) for the US. Unfortunately, rural-
urban differences in social integration and health have not been intensively studied in
Japan. For that reason, I shall use rural-urban residence as an important "test"

variable in this dissertation.

Statement of Questions

Previous studies suggest that having multiple roles is linked to better health of the

elderly by preventing social isolation, increasing social connectedness to society, and

68

buffering stress. Incorporating the life course framework, the elderly who occupy not
only diverse but also dense roles overtime may have better emotional and physical
outcomes than those who do not. These effects are expected to differ by gender,
residence, and age. However, we do not know whether the effect of multiple role
occupancy is significant at any point of time in their life course, regardless of
residence and gender. We do not know whether the effect of gender, residence, and
age remain, even controlling for the effects of multiple roles and socioeconomic
status. We also do not know whether social integration during the adult years such as
marital duration and work duration promote both social integration and health in the
later life course, and whether the number of particular roles matter. To summarize, I
pose three research questions answerable from a national level longitudinal survey

(wave 1 and wave 2) of elderly Japanese people.

Research Question 1: Did the elderly who occupied more roles have a greater
likelihood to be healthy (fewer physical and mental problems) than those with fewer
roles at the first interview? How do gender, age, and residence (rural vs. urban)
influence the well-being of the elderly, controlling for the effects of other variables

including a multiple roles variable and socioeconomic variables?

Research Question 2: Did a particular position/status (e.g., spouse or parent or

grandparent) have a greater effect on health? Did the number of particular roles

matter? How did gender, age, and residence (rural vs. urban) influence the well-

69

being of the elderly alter the effects of other variables, including individual roles and

socioeconomic variables, were controlled?

Research Question 3: Did the elderly, who had diverse roles at the first interview,
have a greater likelihood to remain healthy at the second time of the interview than
those who had fewer roles? Controlling for the effect of other variables, including a
multiple roles variable and socioeconomic variables, did age, gender, and residence

influence health outcomes?

70

CHAPTER 4

METHODOLOGY

Introduction

In the preceding chapter, I asked whether social integration, indicated by
occupying multiple social roles (e.g. being a parent, grandparent, worker), produced
better physical and mental health among elderly Japanese (Research Question 1), and
whether a particular role or numbers of particular roles occupied (e. g. the number of
social groups or social activities the elderly respondents join) have greater effects on
health (Research Question 2). Since I posited that social integration contributes to
good health, then greater integration (seen in the accumulation of more roles) ought
to cause the maintenance of better health (Research Question 3). For all of these
research questions, I asked how effects of social integration differ by gender, age,
and residence (rural vs. urban). Answers to these questions require nationally
representative longitudinal data with multiple indicators of social-role memberships
and of physical and mental conditions. In this chapter, I describe both panels of the
Nihon University Japanese Longitudinal Study of Aging (NUJLSOA), and explain
the appropriateness of this two-year longitudinal data set to provide answers to my
three questions. Then I shall discus the cross-sectional and longitudinal statistical
techniques that I will use to explore answers to my three research questions; and
present the measurements of the independent, dependent, and control variables that

both panels afford.

71

Ja anese Lon 'tudinal Stud of

 

Asia:

The data are from a two-wave panel study of a national sample of 6700 Japanese
elderly aged 65 and over. This longitudinal survey, the Nihon University Japanese
Longitudinal Study of Aging (NUJLSOA), provides panel data based on a nationally
representative sample of the Japanese elderly that takes population size within
regions and prefectures into account. The limitation of this data set is that currently
only two waves (1999 and 2001) are available to the public. However, since few
studies have examined the impact of multiple roles on health in Japan (Takeda et al.
2006), findings from this data set will provide a benchmark for future research on the
dynamic relationship between social integration and health.

A major strength of the data set is that it is the most recent national-level survey
of the Japanese elderly available to the public. Paired sample t-tests examining
changes in the scores for physical, self-rated, and psychological health measures all
showed that there were significant changes in scores between the first wave and the
second wave (Table 1). Hence, its longitudinality allows for analyzing changes in
health status of the elderly, which is another major strength of this data set. In
addition, crucial to the purposes of my dissertation, the questionnaire allows me to
assess the degree of a respondent's social integration according to the diversity of
social positions or statuses the elderly respondent occupied, the duration of roles, and
the numbers of individual roles (e. g., the role depth of parenthood is indicated by the

number of one’s children). For example, I can gauge role diversity from responses to

72

questions on whether the respondent has a spouse, children, grandchildren, a job, or a
membership in an organization outside the family. It is possible to measure the
density of a person’s role as a parent or grandparent from the number of one's
children or grandchildren, respectively. Because experience in a social role may be
needed before its effect on health begins, it is advantageous that the data set allows
me to control marital duration in months and total employment duration in years.
The survey enables me to measure the frequency of contact with kin. The quality of
one's relationship with children is shown by the respondents' reported satisfaction
with amount of contact with them. In sum, the complex measures of social
integration available in Wave 1 make the NUJLSOA an ideal data set for my
dissertation.

Research Questions 1 and 2 ask whether social integration is related to better
health at Wave 1 and to its maintenance until Wave 2. Thus, it is fortuitous that both
waves afford measures of physical ability: Activities of Daily Living (ADLs) and
Instrumental Activities of Daily Living (IADLs), mental health (depression
measures), and self-rated health. As such, it is possible to assess the cross-sectional
association between various dimensions of social integration and diverse indications
of health at Wave 1 and between social integration at Wave 1 and the maintenance or
loss of health over the next two years.

Finally, the NUJLSOA data set allows me to examine how age, gender, and rural-

urban residence affect the relationship between social integration and health. It also

73

allows me to take into account the respondents' educational level and the combined

income of the respondent and spouse.

Method of Sampling

The NUJLSOA survey data were collected using a multistage stratified sampling
method based on three sources: the National Residents Registry System, the list of
eligible voters, and a listing of all households developed by Central Research
Services, Incorporated (CRS). The Registration system in Japan is considered to be
constantly up-dated and universal due to a legal requirement to report any moves to
local authorities within two weeks. The National Residents registry list is the most
credible and updated source of information about respondents. In some
municipalities, that did not allow access to this list, even for research purposes, the
list of voters was used. The list of voters is made based on the National Residents
Registry. Thus, it provides updated and accurate information. If municipalities did
not allow access to either one of the lists, the list of housing units developed by CRS
was used. The list of housing units is regarded as relatively accurate as it has been
maintained for almost 50 years and is regularly used for national samples. CRS is a
private and the most experienced firm in Japan conducting longitudinal surveys. It
was organized in 1954 and has over 600 trained interviewers who do more than 250
surveys with over 300,000 interviews per year. CRS employed 331-trained

interviewers to conduct this survey under the direction of the NUJLSOA.

74

The sampling was done by first stratifying the 47 prefectures throughout Japan
into 11 regions. About 3000 municipalities were then stratified by population size
within regions and prefectures, and 340 primary sampling units (PSUs) were selected
based on a systematic sampling method. In 298 primary sampling units, sample
members were selected from the National Residents Registry of Japan. In another 41
units, sample members were selected from the list of voters, and in one unit, sample
members were selected from the list of housing units developed by CRS. The
population 75 and over was over-sarnpled by a factor of two in order to keep
sufficient numbers of "older" elderly in the longitudinal survey of the elderly. From
each of the 340 PSUs, six to eleven people aged 65 to 74 were selected and eight to
twelve people aged 75 and over were selected. For more information, please go to
NUJLSOA website at www.usc.edu/dept/gero/CBPH/nujlsoal.

Mainly two methods were used to encourage participation. First, newspaper
articles advertised the forthcoming survey as an important task of a major University
research group whose findings would be relevant to older people’s lives. Second,
respondents to the NUJLSOA were given a small incentive worth about US$ 10; two
New Year’s cards plus two other greeting cards were sent, and a summary of the
results from the first interview by CRS. The original sample list of the first interview
selected 6,700 respondents in October, 1999: 2,922 were between the ages of 65-74
and 3,788 were age 75 and above. Of the original list of respondents, 4,640 (69.3
percent) responded to the first wave collected in November and early December,

1999. Three months later, additional attempts were made to interview 710

75

respondents who were either absent, institutionalized, or who did not strongly refuse
to participate in the first interview. This added 357 respondents and brought the
overall response rate to 73.6 percent. Out of the 357 respondents, 49 were
institutionalized during the November interview, and they were given a shorter form
of the interview in March to reduce the burden. Among 4997 respondents, 498 (10
percent) elderly were represented by proxy interviewers. Proxy interviewers were
used in the case of hospitalization (2.2 percent), institutionalization (0.5 percent),
hearing difficulties (2.2 percent), speaking difficulties (0.7 percent), memory 1088
(1.0 percent), dementia (0.9 percent), psychological disorder (0.2 percent), physical
illness or disability (2.2 percent), and other reasons unrelated to health (1.3 percent).
The second wave data were collected in November and December, 2001. The
weighted sample of 4044 participants in the second wave represented a retention rate
of 80.3 percent from the first wave. Three hundred twenty-seven elderly had passed
away, and the rest were lost-to-follow up. Of 4044 participants, 415 (10.3 percent)
elderly were represented by proxy interviewers. Proxy interviewers were again used
in the case of hospitalization (2.4 percent), institutionalization (0.7 percent), hearing
difficulties (1.8 percent), speaking difficulties (0.2 percent), memory 1088 (0.9
percent), dementia (1.2 percent), psychological disorder (0.2 percent), physical
illness or disability (1.7 percent), and other reasons unrelated to health (1.1 percent).
Doubtless, the usage of proxy respondents was an important reason for acceptable

response rates to both waves.

76

Measures

0 Independent variables: Social Integration

Various scholars use role-based integration measures to assess participation in
different types of social relationships (Adelmann 1994a; Brissette, Cohen and
Seeman 2000; Chrouser and Ryff 2006; Kikuzawa 2006; Moen et al. 1989, 1992;
Wethington et al. 2000). In these studies, social integration is conceptualized as "the
number of social roles (types of social relationships) for which respondents report
active participation" (Brissette et al. 2000: 56). These social positions or statuses
become “social roles” when they occur in a system of interaction (e.g. parent,
grandparent, worker). A limitation of my data set is that I have only one indicator of
direct social interaction: the satisfaction of contacts with children. Thus, I use the
terms “social role” and “social position” interchangeably.

Moen et al. (1989) created a variable, ROLES, which consists of a sum of
different roles occupied by respondents at the time of the first interview. They used
six roles: worker, church member, friend, neighbor, relative, and club or organization
member. In Wethington et al.'s (2000) conceptual model, social support was grouped
into two levels by following the number of positions held in a close network (family
and friends) and the number of positions held in a meso-network (institutional
membership and participation). In Kitazawa's study (2006), six roles (spouse, parent,
grandparent, worker, friend, and voluntary organization member) were included as a
measurement of multiple roles. In my research, I measured social integration as sum

of five roles: spouse, parent, grandparent, worker, and social activity memberships.

77

These roles were dummy coded 1 for those currently occupying a particular role and
0 otherwise. Thus, scores of multiple roles ranged from 0 to 5. This is my indicator
of role diversity.

In addition, in this study, the number of each role (e.g. the number of sons,
daughters, the number of social groups and activities participated) was analyzed. 1
classified these into three dimensions: (1) family membership, (2) community
fiiendship, and (3) institutional memberships. Since I represent each dimension by
the total number of roles held, each number represents role depth.

(1) Family Membership : Roles in family were measured by variables indicating
marital status, the number of children, and the number of grandchildren. In the first
wave, 63.5 percent (3171) of respondents were married, and respondents had an
average of 1.15 sons and 1.13 daughters. Those with stepchildren averaged 1.85.
Respondents had 3.77 grandchildren, on average. Nine respondents (less than one
percent) did not provide information about the number of grandchildren. In this case,
since the number of missing cases is very small, the mean number of grandchildren,
3.77, was substituted. This is a method used by statisticians to deal with small
numbers of missing cases, to retain enough cases in conducting categorical analyses.

In addition, respondents were asked: "How many are there in your household,
including yourself?" In both waves, there was an average of three people in the
household of respondents. I used this as another indicator of depth in familial roles.
(2) Communfl' Friendship: Roles in community fiiendships were measured as the

number of social statuses or positions occupied in the following social groups and

78

social activities: (I) respect for the age association, (2) women’s club, (3) senior
citizen’s club, (4) educational or study group, (5) town association, (6) volunteer
work (public service activities), (7) hobby club, and (8) sports club. In the first wave,
on average, respondents participated in one social group or social activity. Forty-two
respondents (less than one percent) did not provide information about the number of
social groups and activities in which they participated. Since the number of rrrissing
cases is small, I substituted the mean (1.08) for missing data.

(3) Institutional Membership: Roles in institutional memberships were measured as
current employment status. The respondents were asked: "Are you currently
working?" In this study, those who were working were scored (1), and those who
were not working were scored (0). In the first wave, 1249 (25.0 percent) respondents

were working.

0 Measuring durations of marriage and work

Previous studies of social integration have ignored how long an incumbent
engaged in a particular role. In this study, two measures of duration, total
employment duration measured in years and marital duration measured in months,
were included in analyses because we do not know whether longer employment and
marital duration enhance an individual’s degree of social integration. Indeed, the
length of time spent in a social role may be more health - protective than the total
number of social statuses or positions occupied. Alternatively, the significance of
being employed or married could depend only on the duration in which a respondent

occupied the role of worker or spouse, respectively.

79

Total lifetime employment was measured in years in the questionnaire. Almost
ten percent (477 respondents) of respondents to the first wave of the survey had
never worked; they were scored zero. Almost five percent (241 respondents) did not
provide information regarding the total years of duration, although they had worked.
In such cases, the mean duration of total employment, 32.56 years, was substituted.

Marital duration was computed as follows. Currently married respondents were
asked the month and the year they married the current spouse. Currently widowed
and divorced respondents were asked about the month and the year their marriage
ended. For married respondents, marital duration in months was computed by
subtracting the year they were born from the year they got married, and multiplying
the remainder by 12 in order to convert this measure into months. Then the month
they were born was subtracted from the month they married, and the remainder was
added to this measure in order to compute the total duration of marriage expressed in
months. Of 3180 married respondents, about seven percent (237 respondents) did not
provide the month they got married. Adopting the assumption frequently made in life
table calculations, I assumed that they married in June (six months). Of 3180 married
respondents, about 4 percent (140 respondents) did not provide the year they got
married. In such cases, I used the average duration of marriage for married
respondents, 555.03 months.

For divorced or widowed respondents, marital duration in months was computed
by subtracting the year they married from the year their marriage ended by death or

separation from their spouse. The remainder was multiplied by 12 in order to convert

80

this measure to months. Then the month they married was subtracted from the month
they were separated from the spouse, and the difference was added to this measure to
compute the total duration of the most recent marriage expressed in months. Of 1723
divorced or widowed respondents, about 15 percent (265 respondents) only provided
the year they got married or were separated from their spouse. For those people, I
added six months to the duration of marriage expressed in months, taking the mid-
year. Of 1723 divorced or widowed respondents, about 17 percent (289 respondents)
did not provide information about the year and the month they got married and
separated from a spouse. In such cases, the mean duration of marriage for divorced
and widowed respondents, 426.46 months, was used. For never-married respondents,
marital duration was 0 month. On average, respondents were married for 500 months
(41.67 years).
0 Measuring the quality of the relationship with their children including

stepchildren

Previous studies of social integration and aging have ignored the quality of
people’s relationships. In this study, the quality of the relationship with children was
included in analyses because the number of children may not be health-enhancing if
communication with them is infrequent and the elderly are unhappy with it.
Respondents were asked whether they were satisfied with the amount of contact with
their children for each child. Scoring “1” for those satisfied with the amount of
contact for each child, I added up these scores. Then the rate of satisfaction with their

contact with children was computed by dividing this sum by the total number of

81

children. Thus, this measure ranged from 0 to 1, and those who did not have children

automatically received “0” for this measure. The mean of this measure was .57.

o Dependent Variables: Health Measures

Physical Health Measures (ADL, IADL): High physical-functional capacity is an
essential component for successful aging (Fujiwara et al. 2003; Rogers, Hummer and
Nam 2000; Rowe and Kahn 1997; Shibata, Sugisawa and Wataname 2001). To
assess the physical functioning of individuals, Katz and his colleagues developed an
Activity of Daily Living (ADL) measure (Rogers et al. 2000), which includes the
following seven items of difficulty: (1) bathing or showering; (2) dressing; (3)
eating; (4) standing up from a bed or chair/sitting down on a chair; (5) walking
around the house; (6) going outside; and (7) toileting. These activities are essential to
daily life. Lawton, Brody, and Fillenbaum developed an Instrumental ADL (IADL)
scale, which reflects activities adopted to one's social environment and to sustain an
independent life in a community (Shibata et al. 2001). In other words, IADLs assess
more complicated activities than ADLs, but may not be required as frequently as
ADLs (Rogers et al. 2000). IADLs include the following seven items of difficulty:
( 1) preparing meals; (2) leaving home to purchase necessary items or medications;
(3) taking care of financial matters; (4) using the telephone; (5) dusting, cleaning,
and other light housework; (6) taking the bus or train to leave home; and (7) taking
medication as prescribed. I scored one for each activity and a maximum score of
seven on any of these ADL or IADL scales denotes greatest functional disability.

The value of Cronbach’s alpha, a coefficient of reliability, among seven ADL items

82

in Wave 1 was .923, among seven IADL items in Wave 1 was .935, and among the
combined 14 ADL and IADL items in Wave 1 was .960.

Generally, studies show that those with greater functional limitations will have a
higher risk of earlier mortality than those with few functional limitations (Rogers et
al. 2000). In fact, this US. study showed that, among younger adults, those with one
or two ADL-IADL limitations were over 40 percent more likely to die younger, and
those with three or more limitations were over twice as likely to die younger
compared to those with no limitation (Rogers et al. 2000). In Japan, several studies
show that disability at any level of functional capacity is likely to shorten life
expectancy (Ishizaki 2004; Shibata et al. 2001). Shibata et al. (2001) carried out
comprehensive studies on aging and problems in the aging society of Koganei City, a
western suburb of Tokyo, from 1976 through 1991, and found that disability in both
ADLs and IADLs increased the odds of death at a younger age. Thus, ADLs and
IADLs are important measures assessing disabilities that lead to a higher risk of
premature mortality. Following Smith and Kington ( 1997), I combined scores on
these two measures to create a maximum score of 14.

In the first wave, 83.1 percent of respondents were independent in all ADL and
IADL items, and 81.1 percent of respondents were independent in all ADL and
IADL items in the second wave. The highly skewed distribution of these measures
was due to the high percentages of respondents with no difficulty performing ADLs
and IADLs; that also is found in other studies targeting the elderly population in the

United States and in Japan (Fredman et al. 1999; Rogers et al. 2000; Schoeni et al.

83

2005; Shibata et al. 2001). I compared the separate scores on ADLs and IADLs and
the composite score for each respondent at Wave 1 and Wave 2 by means of a paired
t-test (Table 1). Table 1 shows that, on average, respondents had less than one
disability out of 14 ADL and IADL items of difficulty at the time of the first wave,
and they had one physical disability out of 14 ADL or IADL items at the time of the
second wave. The statistical significance test showed that the changes in scores
between the two waves were significant (Table 1). However, considerable numbers
of non-responses were found in the ADL and IADL measures, greatly reducing the
sample size in the paired sample t-test (Table 1). To preclude excluding these
respondents from the analyses and to further detect characteristics of these
respondents if they were significantly different from others, I combined the measures
of ADL and IADL items and trichotomized them: (1) no disability, (2) at least one
disability, and (3) no response.

To view changes between two waves, the combined measure of ADL and IADL
items in both waves was categorized into five measures: (1) fewer physical
disabilities than at Wave 1, (2) more physical disabilities than at Wave 1, (3)
continue to have some physical disability, (4) continue to have no physical disability,
and (5) missing cases.

Self-Rated Health: Self-rated health is another indicator associated with health. This
measure has several benefits relative to others. The measure can capture an
individual’s experience of a full array of illness, even symptoms of undiagnosed

disease; and it can represent a person’s complex judgment of health over time more

84

reliably than a medical diagnosis by others (Idler and Benyarnini 1997). Many
studies have found a significant and independent effect of self-rated health on
mortality (Idler and Benyarnini 1997; Ishizaki, Kai and Imanaka 2006). For example,
based on the sample of elderly in Saku City, Nagano prefecture, Japan, Ishizaki et al.
(2006) found, during a six-year interval, that having poor self-rated health was a
significant predictor of mortality.

Questions asking respondents to rate their health differ from study to study
(Ishizaki et al. 2006). In the NUJLSOA survey, an interviewer asked a respondent,
"In general, how would you describe your state of health?" Respondents rated their
overall health using five categories: very healthy; healthier than average; average
health; somewhat unhealthy; and very unhealthy. This measure was scored one to
five (1= very healthy, 2 = healthier than average, 3= average, 4 = somewhat
unhealthy, 5 = very unhealthy). Thus, a higher score indicates poorer self-assessed
health. The paired samples t-tests based on the score ranged from "1" (very healthy)
to "5" (very mrhealthy). Changes in self-rated health between the first and the second
wave showed that there were significant differences in scores (Table 1). In this study,
I grouped these scores into three categories: (1) better than average; (2) average; and
(3) below average. Less than one percent of these samples (47 respondents) did not
provide answers, and they were excluded from the analysis.

To view changes in self-rated health, self-rated health measures in both waves
were compared and further categorized into six groups: (1) Respondents continued to

rate their health better than average; (2) respondents continued to rate their health

85

average; (3) respondents continued to rate their health below average; (4)
respondents rated their health better in wave 2 than wave 1; (5) respondents rated
their health worse in wave 2 than wave 1; and (6) missing cases.

Psychological Health (CES-D Scale): To assess psychological health, a l2-question
version of the Centers for Epidemiologic Studies Depression (CES-D) scale was
used. This measure was developed to tap levels of depressive symptomatology and
was designed to measure multiple dimensions of affective symptomatology
(Callahan and Wolinsky 1994; Takeshita et al. 2002). The CES-D scale usually
contains four clusters of items: depressive affect symptoms ("felt depressed" "felt
sad"), somatic symptoms ("trouble sleeping" "didn't have much appetite", "trouble
feeling motivated", "ordinary things felt troublesome), interpersonal difficulties ("felt
lonely", "felt people were unfriendly" "felt hated by others") and positive affect ("felt
happy", "felt like smiling", "future seemed bright"). In this study, positive affect was
excluded since previous studies suggest that it is not a well-grounded factor for
measming depression in Japan (Kikuzawa 2006; Krause and Liang 1992). The value
of Cronbach’s alpha among nine negative items of CES_D measures in Wave 1 was
.789.

Items in CES-D measures were scaled following Takeshita et al. (2002). No
point, i.e., zero, was given to those who answered "rarely," one point for those who
answered "sometimes", and two points for those who answered "often". A high score
on any of these indicators denotes a greater degree of depression. Scores ranged from

a theoretical low of "0" to a theoretical high of "18." The average score of CES-D in

86

the first wave was 1.432, and 1.452 in the second wave. The paired samples t-tests
comparing changes in CES-D measure between the first and the second wave
showed that there were significant differences in scores (Table 1). However, a
considerable number of non-responses were found in the CES-D measures, which
greatly reduced the sample size in the paired sample t-test (Table 1). To preclude
excluding these respondents from the analysis if they were significantly different
from others, I trichotomized the psychological health measure: (1) zero symptoms of
depression; (2) at least one symptom of depression; and (3) no response.

To view changes in depression scales between waves, the CES-D measures in
both waves were compared and further categorized into five: (1) fewer symptoms of
depression than Wave 1; (2) more symptoms of depression than Wave 1; (3)
continue to have some symptoms of depression; (4) continue to have no symptoms of
depression; and (5) missing cases.

0 Control Variables

The major control variables are age, sex, and residence. In addition, education and
income were included as indicators of socioeconomic variables. Previous studies
found that these variables have positive effects on health in Japan (Fukuda,
Nakamura and Takano 2005; Takagi, Silverstein and Crimmins 2007). Hence,
controlling for these effects refines the process of examining the relationship

between social integration and the well-being of the elderly.

87

4gp: Age was measured as a continuous variable. It ranged 65 years to 99 years in
the first wave. In the first wave, the mean age was 73 years, which became 75 years
in the second wave; it was conducted two years after the first wave in 1999.

flea The variable, sex, was dichotomized. Men were coded (1) and women were
coded (0).

Residence (Rural vs. Urban): The residence measure was obtained in response to the
following question: "What type of community do you currently live in?"
Respondents answered whether they currently live in an urban area ("0") or a rural
area ("1").

Education: The measure of education was obtained in response to the following
question in the first wave. "What was the last educational level completed (includes
withdrawal before completion)?" Seven choices were provided: l= junior high
school (through ninth grade and primary school in the former school system); 2=
high school (through twelfth grade and junior high school, schools for girls, and
teacher schools in the former school system); 3= vocational school; 4 = junior
college (includes technical institutes); 5== university; 6 = graduate school; and 7 =
other. Following Takagi et al. (2007), education was recoded on 4-point scale: (1)
junior high school; (2) high school; (3) vocational school or junior college; and (4) 4
year turiversity or graduate school.

Mp: Toward the end of the survey, respondents were asked: "Approximately
how much do you and your spouse receive as income annually including bonuses

before taxes?" Respondents chose their income from the following ordered

88

categories: (1) less than 500,000 yen; (2) 500,000 yen to 1 million; (3) 1 million to
1.5 million; (4) 1.5 million to 2 million; (5) 2 to 3 million; (6) 3 to 4 million; (7) 4 to
5 million; (8) 5 to 6 million; (9) 6 to 8 million; (10) 8 to 10 million; (11) 10 to 12

million; (12) 12 to 15 million; and (13) more than 15 million.

Statistical Procedures

Looking at descriptive statistics, I saw substantial numbers of "no response" (NA)
and "do not know" (DK) in the health measures, especially for the second wave due
to death and loss to follow-up. In addition, since proxy respondents did not provide
answers for psychological health measures (CES scales), this resulted in a substantial
number of non-responses in both waves. To preclude excluding these populations
from the analyses, and to examine the relationship between multiple roles and health,
multinominal logistic regression was selected to examine the three research
questions. It allowed me to use "no valid response" as one response category.

Multinominal logistic regression is the extension of the logistic regression when
dependent variables are categorical, but not restricted to being dichotomous. For
example, instead of predicting dichotomous outcomes such as being disabled and
non-disabled, I used three outcomes for the first wave: disabled, non-disabled, and
non-response. As discussed above, I used three categories to measure depression at
Wave 1, and five categories to measure changes in depression between 1999 and
2001. Similarly, I used three categories to score self-rated health in 1999 and five to

score changes in this variable between the waves. For a comprehensive discussion of

89

the statistical theory undergirding the multinomial logistic regression and for
examples of its application, the reader is referred to Agresti (1996) and Garson
(2006).

In this chapter, I explained my research methodology, explaining the data set,
measurements, and statistical methodology to answer three research questions raised
in the previous chapter. In the next chapter, I present the findings for my first and

second research questions.

90

CHAPTER 5
RESULT (1)

Cross-Sectional Analyses of the First Wave- Research Questions 1 and 2

This chapter reports findings from cross—sectional analyses of the first wave of the
Nihon University Japanese Longitudinal Survey of Aging (NUJLSOA). The first two
research questions I raised in the third chapter are discussed. For each research
question, I looked at three health outcomes: physical health, self-assessed health, and
psychological health. In addition, for each research question, controlling for such
other variables as socioeconomic indicators, I sought to ascertain whether age,
gender, and residence affected the relationship between social integration and the
health of the elderly.

Multinorrrial logistic regression analyses were used to analyze the NUJLSOA data
set because there were considerable numbers of missing cases due to "No Answer"
(NA) and "Don't Know" (DK) responses on ADL, IADL, and depression scales. This
statistical method allows me to include NA/DK responses into analyses, and that
avoids introducing the biases of selectivity that might arise if those without valid
responses differ systematically from those with valid responses.

Table 2 is illustrative. There we see that the addition of each role in the multiple-
role variable reduced the odds of providing NA/DK responses on ADL or IADL
scales rather than answering they had no physical disability by 15.7 percent (OR =

.843, p<.01). Selection may be at work; those who are disability-free may be selected

91

into more social activities that broaden their number of roles and make them more
conscious of their disability-free status (thus less likely to reply "Don’t Know").
Since women had three times the risk of men to give invalid replies to items on the
ADL or IADL scales rather than to report no disability, it is possible that women
were more challenged in finding time to complete the survey questions on this
section. Thus my study sample would have disproportionately omitted women rather
than men if I had chosen a statistical method forcing me to discard cases with
NA/DK replies to items on the disability scale. In the tables that follow, I report the
odds ratios and standard errors for the NA/DK response; but for the purpose of
parsimony, I discuss only the portions of the tables that provide answers to my

research questions.

Research Question 1
Research Question I : Did the elderly who had more roles have greater odds of being

healthy (fewer physical and mental problems) than those who had fewer roles at the
first interview? How did gender, age, and residence (rural vs. urban) influence the
well-being of the elderly, controlling for the efihcts of other variables including a
multiple roles variable and socioeconomic variables?
0 1 — A: Physical Disability (Table 2)

Table 2 summarizes the results of a multinomial logistic regression that compares
respondents in three response categories: (1) those with one or more physical

disabilities on ADL or IADL scales; (2) respondents with at least one of the answers

92

on ADL or IADL scales missing; and (3) respondents with no physical disability on
ADL and IADL scales. The last category is treated as the reference category.

As expected, Table 2 showed that controlling for the effects of other variables, the
addition of each role membership counted in a multiple roles variable reduced the
odds of worsening physical disability by 45.4 percent (OR = .546, p < .001).
Holding the effect of other variables constant, the elderly who occupied diverse roles
had a lower risk of physical disabilities than those who had fewer roles. This result
implies that a positive effect of the multiple roles variable on physical health is
significant at any age in the later life course, and it is significant regardless of gender
and residence. This result supports the role enhancement perspective which argues
that holding multiple roles leads to better physical well-being of the elderly and
refutes the notion that multiple roles strain older people and cause deterioration in
their health.

Every additional member living in the same household increased the odds of
physical disability by 8.2 percent (OR= 1.082, p<.01). Brown et al. (2002) reported,
based on their research of the elderly in Japan, that physical and mental conditions
were measures of transitions in living arrangement. In other words, poor physical
health triggers the creation of Daikazoku, a large extended family tied together
through blood relationships, marriage, or adoption. Influenced by the ie tradition,
dramatized by the media, the word often accompanies the nostalgic image that the
elderly are respected and traditional values are passed down from generation to

generation — virtues that the nuclear family does not offer. My findings support

93

Brown et al. (2002) and suggest that physical disabilities propel elderly people into
living arrangements with several other relatives who can provide assistance rather
than the traditional continuation of Daikazoku over generations.

In addition, every month of increase in marital duration slightly increased the
odds of physical disability (OR = 1.001, p<.001), and each additional year of age
increased the odds of physical disability by 11 percent (OR = 1.110, p<.001).
Umberson et al. (2005) claimed that marital duration presumes that change in marital
quality occurs because of increasing time spent married, possibly due to boredom or
diminished compatibility. Thus, a longer marital duration may indicate poorer
marital quality rather than cumulated emotional support ("Aun no kokyii" ie. a highly
communicative relationship developed over time as if two of married people were
melted into one). However, since the net effects of age and marital duration are
difficult to separate, the negative effect of marital duration could mean that older
elderly are more likely to have physical disabilities than younger elderly. Failure to
include variables measuring marital quality prevent investigation of how marital
quality and marital duration influence physical disability of the elderly in Japan. This
question must await future research.

As expected, Table 2 shows that, controlling for other variables, each higher level
of education reduced the odds of disability by 19 percent (OR = .81, p< .01). This
supports previous studies that showed a positive association between education and
health (Ross and Wu 1995). The result suggests that the well educated are more

likely to avoid disability than the less educated.

94

Finally, Table 2 shows that, compared to rural residents, urban residents were
28.5 percent less likely to have any disability (OR = .715, p<.01) rather than no
disability. This result supports the findings of Ogawa (2004) who used the same data
set but different statistical techniques to analyze rural-urban differentials in health
conditions. Ogawa (2004) found that the rural elderly lived longer than the urban
elderly, but the urban elderly enjoyed a longer duration of being free from physical
disabilities than the rural elderly. In Japan, the urban elderly are more likely to be
influenced by institutional factors such as retirement age because they are more
likely to be in a highly organized labor market than are rural elderly (Ogawa 2004).
In other words, the urban elderly are more likely to retire before their health
deteriorates and may have more time to detect and recover from their health
problems in early stages by going to see doctors than do the rural elderly. The reason
could also be differences in the types of jobs the elderly hold in rural areas versus
urban areas, since descriptive statistics showed that the rural elderly were more likely
to be involved in jobs related to agriculture, forestry, or fishing than the urban
elderly. These jobs not only encourage the rural elderly to work until their health
deteriorates, but also may increase their risk of musculoskeletal injuries.

Japan began a universal health insurance system and the distribution of pensions
beginning in April 1961, thus all Japanese citizens receive pensions. There are three
types of pensions: (1) national pension (kokumin nenkin) for the self-employed; (2)
employee pensions (kosei nenkin) for salaried persons; and (3) mutual aid pensions

(kyosai nenkin) for civil servants (Ministry of Foreign Affairs 2007). A two-tier

95

pension system was established beginning in 1986 and the entire population is
eligible to receive a national pension, to which employees' pensions and mutual aid
pensions are added for people who are eligible (Ministry of Foreign Affairs 2007).
Therefore, not only because the urban elderly are more likely to belong to a highly
organized labor market, but also because they are more likely to receive second-tier
payments, they are more likely to retire early, before their health conditions
deteriorate, than are rural elderly.

Therefore, answering the first research question, the elderly who had more roles
had greater odds of being physically healthy than those who had fewer roles at the
time of the first interview. Both age and residence influenced the well-being of the
elderly, controlling for the effects of other variables including a multiple roles
variable and socioeconomic variables. Gender, however, did not show any
significant effect on physical disability of the Japanese elderly (Table 2).

o l - B: Self-Assessed Health (Table 3)

Table 3 summarizes the results of a multinomial logistic regression on self-rated
health. It compares respondents who rated their health "better than average",
respondents who rated their health "worse than average", and respondents who rated
their health "average". The last group is treated as the reference category. For self-
rated health outcomes, only .9 percent (47 respondents) provided NA/DK responses.
Thus these respondents were excluded from the analyses.

Ceteris paribus, the addition of each role membership counted in the multiple

roles variable increased the odds of rating one's health "better than average" rather

96

than "average" by 16.6 percent (OR = 1.166, p<.01). Moreover, the addition of each
role membership counted in the multiple roles variable reduced the odds of people
rating their health "worse than average" rater than "average" by 30.3 percent (OR =
.697, p<.001). These results indicate that, occupying multiple roles not only
increased the likelihood of people rating their health "better than average", but it also
reduced the likelihood of people rating their health "worse than average". Such a
positive effect of multiple roles on self-rated health was significant at any age in the
later life course and it was significant regardless of gender and residence. Therefore,
the findings support the role enhancement perspective and refute the roles strain
perspective. Holding multiple roles leads to better self-rated health by the elderly, as
well as to less disability of the elderly.

Ceteris paribus, every month of marital duration slightly lowered the odds of
people rating their health "better than average" rather than "average" (OR = .999,
p<.01) and raised the odds of rating their health "worse than average" rather than
"average" (OR = 1.001, p<.05). As expected, each additional year of age reduced the
odds of individuals rating their health "better than average" rather than "average" by
2.1 percent (OR = .979, p<.01), and increased the odds of rating their health "worse
than average" rather than "average" by 1.3 percent (OR = 1.013, p<.05). Similar
effects of age and marital duration were also found for physical disability (Table 2).
Since age and marital duration are two different ways of measuring physiological
aging, these outcomes are not surprising. Apparently, however, a long marriage, per

8c, is not a good solid indicator to measure marital satisfaction. Future studies need

97

to include variables that measure marital quality in order to look at the relationships
among marital quality, marital duration, and the well-being of the elderly over the
life course.

While education was unrelated to self-related health, income was related. A
higher level of income increased the odds of rating one's health "better than average"
rather than "average" by 4.2 percent (OR = 1.042, p <.05) and reduced the odds of
people rating their health "worse than average" rather than "average" by 5.1 percent
(OR = .949, p <.05). These results indicate that, although the positive effect of the
multiple-roles variable on self-rated health was stronger than that of income, the
latter variable played an important role in providing greater resources and financial
security in the later life course and boosted the self-assessed health of the elderly.
Contrary to my expectations, neither being a rural resident nor a woman improved
one's perception of being healthy. Since self-rated health requires an evaluation of
both physical and mental well-being, it is possible that greater access to incomes for
elderly urban women advantaged them over their rural counterparts.

o l — C: Psychological Health (Table 4)

Table 4 reports the multinomial logistic regression of predictor variables upon the
depression scale scores, an indicator of psychological health. Ceteris paribus, for
every social-role membership, the odds of having any symptom of depression fell
11.1 percent (OR = .889, p < .001) below that of having no symptoms. Therefore, the
result supports the role enhancement perspective that multiple roles lead to a greater

level of subjective health of the elderly. This result, consistent with those for

98

physical disability (Table 2) and self-rated health (Table 3), provides strong evidence
that having multiple social roles promotes the health of elderly Japanese along
several dimensions of health and refutes the notion that multiple roles strain older
people and cause deterioration in their health.

The longer the marital duration, the greater the odds of having some symptom of
depression rather than no symptoms (OR = 1.001 , p< .05). This unfavorable effect of
marital duration was also found for physical disability (Table 2) and self-rated health
(Table 3). A further investigation using a different data set that includes rich
information about marital quality is needed to further examine why longer marital
duration negatively influences the well-being of the elderly. It could be due to
boredom or due to the positive association between age and the duration of marriage.

The number of people living with the respondent reduced the latter's risk of
reporting at least one depressive symptom rather than no symptoms (OR = .92,
p<.001). This result indicates that, although physical disability increases the chance
that the elderly will live with other family members (Table 2), the elderly are less
likely to suffer depression in this context. This result indicates that living with other
family members prevents social isolation and boosts the spirits of disabled older
Japanese people.

As expected, Table 4 shows that each higher income level reduced the odds of
having some symptoms of depression by 6.9 percent (OR = .931, p <.001). Similarly,

a salutary effect of income on self-rated health was reported in Table 3. It appears

99

that a higher income provides financial security, which in turn positively influences
health outcomes.

In contrast to its relationships to physical health outcomes (Table 2) and self-rated
health outcomes (Table 3), gender showed significant effects on psychological health
outcomes. To wit, women were 23.9 percent less likely than men to have any
symptoms of depression (OR = .761, p <.01) rather than none. This finding supports
those of previous studies on aging and the family which found that women, but not
men, are better at maintaining and extending social networks (Antonucci and
Akiyarna 1995). Thus, women are less likely than men to feel isolated in the later life
course. This finding also refutes Durkheim's gendered assumption that women are
less involved in public life than men and thus less likely to obtain either the benefits
or the costs of that involvement. The result from this study also reflects the fact that
women tend to apply for and receive more social supports from the government and
their relatives than men do in Japan (Ikeda et al. 2007). A socially constructed and
gendered life course seems to protect the mental health of the female elderly more
than the male elderly in Japan.

Finally, urban residence had adverse effects on psychological health outcomes
(Table 4). Compared with the rural elderly, the urban elderly were 28.9 percent more
likely to have some symptoms of depression (OR = 1.289, p<.001) rather than no
symptoms. This result supports the finding of Fukuda et al. (2005) that urban
residence offers a less supportive network than rural residence in Japan. Perhaps

rural people are more willing to become familiar and helpful neighbors. That would

100

help rural elderly avoid social isolation and reduce stress in their daily lives. Another
reason would be that intergenerational households are more common in the more
traditional and less developed areas of Japan. Thus, the rural elderly are more likely
to be in contexts where they can gain a variety of social statuses or positions within

and beyond the family than the urban elderly.

Research Question 2

Research Question 2: Did particular roles have a greater effect on health? Did the
number of particular roles matter? How did gender, age, and residence (rural vs.
urban) influence the well-being of the elderly, ceteris paribus?

To answer the second research question, rather than including one variable that
measured the effect of multiple roles, I included individual roles in the analyses. This
way, I could analyze whether holding a greater number of the same social position
(e. g. parent to a larger number of sons) matters in the degree of social integration and
influences the well-being of the elderly. For example, instead of including dummy
variables showing whether the respondent was a parent or grandparent, I included the
number of sons, daughters, stepchildren, and grandchildren a person had as separate
predictors of the health measures. In addition, each analysis was divided into four
models (See Tables 5-A, S-B, 6-A, 6-B, 7-A, and 7-B). The first model includes two
categorical variables (marital status, work status); the second model adds residence
(rural vs. urban); the third model adds gender to Model 1; and in the fourth model,

both residence and gender were added to Model 1. A weakness of multinomial

101

logistic regression is that cells in the crosstab that look at the relationship between
categorical independent variables and a categorical dependent variable should not be
small. Since the last model includes all of four categorical variables, one or two cells
were not zero cells, but contained fewer cases than five, as indicated in notes to
Table 5-A, Table 5-B, Table 7-A, and Table 7-B. Therefore, categorical variables
were not entered all at once, but were entered one-by—one into analyses.

0 2 — A: Physical Disability (Table 5-A)

Compared to respondents who were working at the time of the interview,
respondents who were not working had three times the odds of having at least one
disability on ADL or IADL scales rather than no physical disability (OR = 3.0,
p<.001; Table 5-A). Controlling for other variables, additional participation in social
activities and social groups reduced the risk of physical disability by 50 percent (OR
= .5, p<.001). This result is consistent with that of a US. study which found that
being a member of a social group or organization appeared to be especially salutary
(Moen et al. 1992). These results imply that having extrafarnial roles is important.
Perhaps healthy elderly people are more likely to remain in the labor force and to
participate in more social activities and social groups than are unhealthy people. As
Moen et al. (1992) claim, the pathways to successful aging appear to be a cumulative
process which reflects choices and experiences throughout early and middle
adulthood.

Unexpectedly, the data in Table S—A showed that having more daughters (not

sons) increased the risk of physical disability by 16 percent (OR = 1.16, p<.05).

102

However, I argue that this result may reflect a statistical interaction between the
number of daughters and rural residence. For example, running the independent
sample t-tests among respondents who have at least one child, I found no significant
difference in the average number of children between those who had daughters only
and sons only (1.91 daughters for respondents with no sons vs. 1.82 sons for
respondents with no daughters). Restricting the analysis to urban residents, there was
also no significant difference between the two groups (1.85 daughters vs. 1.77 sons).
However, restricting the analysis to rural residents, there was a significant difference
between the two groups (2.10 daughters vs. 1.88 sons), which implies that rural
respondents had a longer childbearing period than urban respondents. The older the
age at childbirth, the more likely the child is to be a daughter (Shryock and Siegel
1976). Thus, the total number of daughters may be a proxy for late childbearing and
the social disadvantages that may be associated with it in rural areas. Further studies
looking at rural-urban differentials in birth intervals are necessary to examine the
possibility of biological factors.

Another possible interpretation of the association between residence, the number
of own daughters, and physical health is that, reflecting the ie system, the daughter is
the one who will leave her ie and become a member of her husband's ie. Because of
such a patriarchal tradition persistently influencing Japanese society, parents of
daughters are often cautious not to let their own daughters feel embarrassed in the
new ie and to be welcomed in the new ie. For example, parents of daughters often

buy furniture and formal clothes for their daughters when they get married. In

103

various regions, parents of daughters prepare for a traditional festival of their
grandchildren by purchasing dolls and kimonos. In Japan, parents of daughters take
care of them when they give birth. Daughters spend a month at their parents' house
before and after they give birth. Since sons tend to remain a member of the original
ie, parents of sons may not need to worry about such a matter. A very popular
Japanese television family drama called, Watarusekenwa oni bakari, is a story of
elderly parents worrying about their five daughters in their new ie and supporting
them by listening to their daughters conflicts and soothing them. Such worries often
caused the old parents depression and physical tiredness. This result opens doors for
future studies since many studies have until now focused on the tension between a
mother-in-law and a daughter-in-law but have ignored the relationship between the
number of daughters (not daughters-in-law) and the health status of the elderly.
Especially needed are qualitative studies that look at the quality of the relationship
between elderly parents and the sex of children to find how the number of sons or of
daughters benefits or disadvantages the well-being of the elderly.

Greater satisfaction with the level of contact with their children, including
stepchildren, reduced the risk of physical disability by 50 percent (OR = .50, p<.001)
relative to reporting no disability. On the other hand, having a greater number of
members living in the same household did not show a significant effect on the
physical well-being of the elderly (Table 5-A) Descriptive statistics showed that the
association between respondents' satisfaction with the level of contact with their

children and the number of household members was significant but negative

104

(Pearson's R = -.294). This indicates that the positive effect of the level of contact
with their children on the health of the elderly is not due to living with more people
in the same household. Living with more people could cause a financial burden,
whereas being satisfied with the amount of the contact with children positively
influences health outcomes regardless of coresidence. In addition, having a greater
number of sons and stepchildren did not show a significant effect on the physical
well-being of the elderly (Table 5-A). Therefore, looking at individual roles, and
controlling for the effects of other variables, I find that the quality of the relationship
with children is more important to maintaining the physical well-being of the elderly
in the later life course than how many children they have and how many people are
living in the same household.

Looking at other variables, the data in Table 4-A showed that additional age
measured in years increased the odds of physical disability by 11 percent (OR =
1.11, p<.001) and a higher education level reduced the odds of physical disability by
15 percent (OR = .85, p<.05). These results indicate that the chance of morbidity
increased as one became older. However, this probability was reduced or postponed
if one had a higher level of education. As previous studies claim, education
positively influences health outcomes by increasing knowledge to avoid unhealthy
behaviors and to live healthy lifestyles.

o 2 - B : Self-Rated Health (Table 6-A, Table 6-B)
Table 6-A and Table 6-B summarizes the result of the multinomial logistic

regression on self-rated health. Table 6-A compared respondents who rated their

105

health "better than average" to respondents who rated their health "average" (the
reference group). Table 6-B compares respondents who rated their health "worse
than average" and those who rated their health "average", the reference group.

Table 6-A and Table 6—B show that, among five roles (spouse, parent,
grandparent, worker, social activity member), only two variables - work status and
the number of social activities and social groups - were significant. Table 6-A shows
that, controlling for the effect of other variables, compared to respondents who were
working at the time of the interview, respondents who were not working were about
35 percent less likely to rate their health "better than average" rather than
"average"(OR = .65, p<.001). Table 6-B shows that, holding other variables constant,
compared to respondents who were working at the time of the first wave of
interviews, respondents who were not working had 2.3 times the same odds of rating
their health "worse than average" rather than "average" (OR = 2.3, p<.001). Table 6—
A shows that additional participation in social activities and social groups greatly
increased the odds of rating their health "better than average" rather than "average"
by 11 percent (OR = 1.113, p<.001). Table 6—B shows that additional participation in
social activities and social groups reduced the odds of people rating their health
"worse than average" rather than "average" by 23 percent (OR = .77, p<.001).
Similar results were also found for physical disability outcomes (Table 5-A)
Therefore, holding other variables constant, not only engaging in work roles but also

engaging in diverse community roles beyond the family are important for the elderly

106

to rate their health "better than average" and to avoid rating it "worse than average"
rather than "average".

Table 6-B shows that a greater level of satisfaction with the amount of contact
with children, including stepchildren, reduced the odds of rating one's health "worse
than average" rather than "average" by 32 percent (OR = .68, p<.01). Moreover,
Table 6-B shows each additional member living in the same household reduced the
odds of rating one's health "worse than average" rather than "average" by 4 percent
(OR = .96, p<.01). These results indicate that, living with more family members and
having good relationships with children are important factors that prevent the elderly
fi'om rating their health "worse than average". Therefore, these results suggest that,
not only holding individual roles beyond family membership, but also maintaining
frequent and satisfactory family relations contributed to better self-rated health by
the elderly than by those holding fewer individual roles beyond family membership.

As expected, Table 6-A shows that age had a significant effect on self-rated
health. Controlling for the effect of other variables, each additional year in age
reduced the odds of rating one's health "better than average" rather than "average" by
2 percent (OR = .976, p<.01). Because Table 5-A shows a negative effect of age on
physical disability of the elderly, it seems that the chance of rating oneself above
average in health and disability-free declines with advancing age, possibility due to
the biological consequences of aging.

As expected, Table 6-B shows that income had a significant effect on self-rated

health. A higher level of income reduced the odds of rating one's health "worse than

107

average" over "average" by 4.1 percent (OR = .959, p<.05). This result indicates that
financial resources may be used to postpone physical deterioration due to aging.

Finally, Table 6-A shows that, compared to male elderly respondents, female
elderly respondents were about 24 percent more likely to rate their health "better
than average" rather than "average" (OR = 1.24, p<.05). This result reflects the fact
that women tend to receive more social support from the government and their
relatives than men do in Japan (Ikeda et al. 2007). In other words, gendered later life
course benefits women more than men in Japan, which may contribute to the longer
life expectancy of women in Japan.

Although rural people were more likely than urbanites to report being disabled
rather than disability-free (Table 2), rural people did not differ from urbanites in self-
rated health. Since self-assessed health reflects both physical and psychological
health conditions, the negative effect of rural context on physical well-being and the
positive effect of rural context on mental health may cancel the total effect of rurality
on self-rated health.

0 3 — A: Psychological Health (Table 7-A)

I shall now interpret the results in the multinomial logistic regression comparing
the odds of reporting any symptoms of depression to people reporting that they have
no symptoms, the reference category (Table 7-A). Among five roles (spouse, parent,
grandparent, worker, and social activity member), two variables — spouse and worker
status - showed a significant effect and in the direction I expected. Table 7-A shows,

controlling for the effect of other variables, compared to married respondents,

108

respondents who were not married (divorced, widowed, never-married) were about
50 percent more likely to have at least one symptom of depression rather than no
symptoms (OR = 1.5, p<.001). Compared to respondents who were working at the
time of the interview, respondents who were not working were about 35 percent
more likely to have at least one symptom of depression rather than no symptom (OR
= 1.35, p<.01). In contrast to physical health and self-rated health outcomes (Table
5-A and Table 6-A), greater participation in community activities and social groups
did not have a significant positive effect on depression. This result indicates that
having a spouse and a job were important in being physically and mentally healthy,
perhaps through reducing social isolation.

As expected, a higher level of income reduced the risk of depression by five
percent (OR = .95, p<.01; Table 7-A). On the other hand, a higher level of education
increased the risk of depression by 11 percent (OR = 1.11, p<.05; Table 7-A).
Compared with Table S-A, education had a positive influence on physical well-being
on the elderly, but it had a negative effect on psychological well-being of the elderly
(Table 7-A). These results suggest that, socioeconomic variables have more complex
effects on the well-being of the elderly than I expected. The possible interpretation is
that higher education increases the experience of felt responsibilities and burdens of
the added number of social roles characteristic of better educated people in Japan.

Each additional member living in the same household reduced the odds of
depression by nine percent (Table 7-A). Compared to rural respondents, urban

respondents were about 30 percent more likely to have at least one symptom of

109

depression rather than no symptom (OR =l.302, p<.001; Table 7-A). These results
suggest that socially isolated elderly in the household and in the community were
more likely to feel depressed than the elderly surrounded by more people they knew
in the household and in the community.

Finally, a longer duration of marriage increased the odds of depression by 0.1
percent (OR = 1.001, p<.001; Table 7-A). Since variables regarding marital quality
were not included in this data set, my ability to investigate further how marital
duration is related to marital quality is limited. However, it is interesting to notice
that being married lowered the odds of having some symptoms of depression rather
than no symptom whereas being married longer slightly increased the odds of having
some symptoms of depression rather than having no symptom. It is possible that
longer marriage increased boredom with the spouse. Although the focus on this
research is to look at effects of multiple roles on health, the exclusion of variables

measuring marital quality is a limitation of this data set.

Summar_'y

Cross-sectional analyses of the first wave of the NUJLSOA supported the role
enhancement perspective, but not the role strain perspective. The answer to the first
research question is, the elderly who had more roles measured by a multiple-role
variable had a greater likelihood to be healthy than those who had few roles at the
time of the first interview. As previous studies of the role enhancement perspective

suggest, engaging in diverse roles is associated with higher levels of physical and

110

emotional health because the accumulation of social identities or roles benefits
individuals.

Controlling for other variables including a multiple roles variable and
socioeconomic variables, residence had a significant effect on physical and
psychological health, but not on self-rated health. It is interesting to notice that urban
residence reduced the risk of physical disability whereas it increased the risk of
depression. Controlling for other variables, age had a significant negative effect on
physical health and self-rated health, but not on psychological health. Controlling for
other variables, gender only had a significant effect on psychological health.
Therefore, age, gender, and residence all had significant effects on health outcomes
controlling for the effect of other variables including a multiple role variable and
socioeconomic variables. However, how these three variables influenced the health
outcomes was more complicated than previously studied.

Looking at individual roles, I see that not all individual roles had a significant
influence on health. Although diversity of roles contributed to making the elderly
feel more valuable and secure, role density - having more sons, daughters,
stepchildren, and grandchildren — did not positively influence the well-being of the
elderly. This result suggests that, as the role enhancement perspective emphasizes,
holding diverse roles benefits the health of individuals by enhancing individual
resources, social connections, prestige, emotional gratification, and social identity
(Barnett and Marshall 2001; Chrouser and Ryff 2006). In terms of participation in

social groups and social activities, role density had a significant positive influence on

111

physical health and self-assessed health outcomes, but did not have a similar effect
on psychological health outcomes. These results suggest that, rather than role depth,
role breadth is more important for the elderly in Japan. For reasons that deserve
exploration in future research, such breadth provides them with a greater support
network, even if their relationships become stressful (Chrouser and Ryff 2006;

Kikuzawa 2000).

112

CHAPTER 6
RESULT (2)

Longitudinal Analyses of NUJLSOA Data Set - Research Question 3

This chapter analyzes the data to provide an answer to my third research question
by looking at changes in health conditions between the first wave and the second
wave of the Nihon University Japanese Longitudinal Survey of Aging (NUJLSOA)
data set. First, the effects of independent variables including a multiple-roles
variable, age, gender, residence, and socioeconomic status on changes in three types
of health outcomes — physical health, self-rated health, and psychological health —
were analyzed. For statistical analyses, multinomial logistic regressions were used
because there were sizable numbers of missing cases due to "No Answer" (NA) and
"Don't Know" (DK) responses on health outcomes, death, or the inability to locate
the respondents at the second wave.

Table 8 is illustrative. There we see that respondents for whom I could not
calculate a disability score at Wave 2 differed from those who were disability-free at
the time of both waves: the former were more than twice as likely to be women.
Perhaps because men have an earlier average age at death than women, men may be
more likely to avoid answering “don’t know” to questions on disability. Since
women had 2.3 times the risk of men not to have valid replies to items on the ADL
or IADL scales rather than to report freedom from disability at the time of both

Waves, my study sample would have disproportionately omitted women rather than

113

men if I had chosen a statistical method forcing me to discard cases with NA/DK
replies. Therefore, the results of this chapter not only contribute to our understanding
of social integration and health over time, but also minimize the biases that would
beset my conclusions because of sample attrition. The longitudinal analyses below
are initial steps to understand how social integration might retard age-related
declines in health among elderly Japanese. In the tables that follow, I report the odds
ratios and standard errors for the NA/DK response; but in the interest of parsimony, I
discuss only the portions of the tables that provide answers to my third set of
research questions.

The limitation of the multinomial logistic regressions is that it requires having at
least five observations in each cell of the cross-tabulations among categorical
dependent and independent variables. Five is a rule of thumb often used to validate
cross-tabulation analyses. I collapsed categories of independent and dependent
variables as necessary to assure a large enough number of cases in each cell to
produce statistical power. However, there are some subcategories of change in health
conditions that have fewer than five observations due to the short time interval (two
years) between the waves. Future waves will increase the duration of observation
since Wave 1 and remedy the problem of sparse observations in a few cells. Until
future waves become available in the public domain, I present what we can learn

from Wave 2.

114

Research Question 3

Research Question 3 .' Did the elderly, who had diverse roles at the first interview,
have a greater likelihood to remain healthy at the second interview? Controlling for
the eflect of other variables, including a multiple roles variable and socioeconomic
variables, did age, gender, and residence influence health outcomes?

0 3 - A: Physical Disability (Table 8)

Table 8 presents the result of the multinomial logistic regression, examining the
factors that change the odds that a respondent will transition from having at least one
disability to freedom from disability at Wave 2 [first three columns; being disability-
free to disabled at Wave 2 (cols. 4-6); from having some disability at the time of both
waves (cols. 7-9); and from having incalculable disability due to missing data at one
or both waves (cols. 10-12); versus being disability-free at the time of both waves
(the reference group)].

Greater integration, conceptualized as social positions in a system of interaction
and measured as a sum of five roles (spouse, parent, grandparent, worker, and
voluntary organization member), contributed to the maintenance of better physical
health. Each role occupied in a different category of social-group membership had no
effect on the reduction of disability between the waves (OR = .793, not significant).
Yet, the addition of every role membership, i.e., increasing role diversity, reduced
the odds of the onset of physical disability by the second wave rather than
maintaining no physical disability at both waves by 17.9 percent (OR = .821 , p<.05).

Ceteris paribus, each role membership at Wave 1 reduced the risk of continuing to

115

have some physical disability at both waves rather than none at both waves by 51.1
percent (OR = .489, p<.001). These results support the role enhancement perspective
and the life course perspective that occupancy of diverse roles (role breadth) delays
the odds of physical disability. Holding diverse roles not only enables the elderly to
be healthy, but enables the elderly to be continuously healthy. These results mean
that the answer to Research Question 3 i8 affirrnative.

Looking at other variables, age showed a significant effect on the maintenance or
change in physical health outcomes. Table 8 shows that, controlling for other
variables, the older respondents were, the more likely they were to have some
physical disability at the second wave rather than to have no physical disability at the
time of both waves (OR = 1.107, p<.001). Controlling for other variables, the older
the respondents were, the more likely they were to have some physical disability
over time rather than to have no physical disability over time (OR = 1.127, p<.001).
Table 8 also shows that, controlling for the effect of other variables, additional age
measured in years increased the odds of going from disabled to disability-free
between the first and the second wave (OR = 1.059, p<.01). This result indicates
that, even among elderly people, recovery from disability is possible and is likely a
reason why Japan is the most long-lived society in the world. This result refutes the
stereotypical simple assumption that age increases morbidity.

Table 8 shows that a higher level of education reduced the risk of progressing
from being disability-free at Wave 1 to having some disability at Wave 2 rather than

remaining disability-free (OR = .751, p<.05). Table 8 also shows that a higher level

116

of education reduced the risk of recovering completely from disability rather than
having no physical disability at both waves (OR = .618, p<.01). These results
indicate that, a higher education level contributed to a person remaining physically
healthy over time, but it also reduced the risk of having physical disability at the time
of the second wave. Education enables one to acquire, understand, and act upon
health-related information.

Another socioeconomic variable, income, also showed a significant effect on
physical health outcome. Table 8 shows that, controlling for other variables, a higher
income level reduced the risk of having some physical disability rather than having
no physical disability by 8 percent (OR = .917, p<.05). As expected, socioeconomic
variables positively contributed to the maintenance or improvement of physical well-
being of the elderly in Japan.

A greater satisfaction with the level of contact with children, including
stepchildren, also positively contributed to the physical well-being of the elderly
over time. A higher level of satisfaction with the frequency of contact with children
reduced the odds of progressing from a disability-free to a disabled status rather than
staying disability-free at the time of both waves by 37.1 percent (OR = .629, p<.05).
However, a longer duration of marriage slightly increased the odds of having some
physical disability rather than having no disability over time (OR = 1.001, p<.01).
Longer marriage could induce either boredom or happiness; thus additional variables
are required to examine the relationship between marital quality and health

conditions over time. Finally, duration of working was unrelated to changes in health

117

between the waves. This result suggests that holding diverse social roles is more
important than holding one social position for a longer period of time.

Compared with rural residence, urban residence reduced the odds of increasing
physical disability at the second wave rather than having no physical disability at the
time of both waves by 28.1 percent (OR = .719, p<.05; Table 8). In other words, the
urban elderly were more likely to remain healthy over time than were the rural
elderly. This result is consistent with the cross-sectional result which showed that
urban residents were less likely to have physical disability compared to rural
residents. The reasons for this finding include differences in types of jobs and
pension systems between rural areas and urban areas. Living in urban areas may
allow the elderly to have more access to formal health-care services that can retard
progression into disability. Recently, an article appeared in the New York Times
which indicated that urban centers like Tokyo are thriving, whereas rural regions are
experiencing empty downtowns and factories due to a changing economic
environment such as globalization of the market and the government's deep cuts in
public work projects to liberate the private sector and shrink the government (F ackler
2007). Depopulation in rural areas has caused problems including unequal access to
sufficient firnds to provide needed health and social services, educational
opportunities, and the lack of good jobs (Thompson 2003). Such problematic
population politics may have disadvantaged the elderly in nrral areas. I used the
information about residence of respondents from the first wave. Since only 2.4

percent (81 elderly respondents) changed their residence between waves, we must

118

wait for future waves to accumulate enough cases to examine how changes in
residence influence the well-being of the elderly.
0 3 — B: Self-Rated Health (Table 9)

Table 9 summarizes the results of the multinomial logistic regression, examining
if diverse social integration at Wave 1 promoted improved self-rated health at the
second Wave (cols. 7-9) or its deterioration (cols. 10-12) rather than maintenance of
better-than-average self-rated health (the reference group).

As expected, greater integration, seen in the accumulation of diverse roles,
contributed to the maintenance of better self-assessed health. Table 9 shows that,
having diverse roles reduced the odds of people rating their health average at both
waves rather than rating their health better than average at both waves (the reference
category) by 23.6 percent (OR = .764, p<.001). Having diverse roles also reduced the
odds of rating their health worse than average at both waves rather than rating their
health better than average at both waves (OR =.557, p<.001). Since multiple roles
contributed to the maintenance of better self-rated health at the time of both waves,
having diverse roles also reduced the odds of rating their health better at the time of
the second wave rather than rating their health better than average at the time of both
waves by 24.6 percent (OR = .754, p<.001). It also reduced the odds of people rating
their health better than average at the first wave rather than better than average at
both waves by 16.8 percent (OR = .832, p<.01). These results support the role

enhancement perspective and the life course perspective that diverse role occupancy

119

helps people maintain better health conditions and delays changes in their life
trajectory of health.

Looking at other variables, Table 9 shows that a higher education level had a
similar effect with the multiple roles variable, although the positive effect of the
multiple roles variable on self-rated health was stronger than a higher level of
education. A higher level of education reduced the odds of rating their health worse
than average at both waves rather than rating their health better than average at both
waves by 19.4 percent (OR = .806, p<.01). It also reduced the odds of rating their
health average at both waves rather than rating their health better than average at
both waves by 19.7 percent (OR = .803, p<.01). Since higher education level
contributed to the maintenance of better self-rated health over time, the elderly with
higher education were less likely to change their self-rated health over time. A higher
education level reduced the odds of rating their health better at the second wave
rather than better than average at both waves by 12.2 percent (OR = .878, p<.05). It
also reduced the odds of rating their health better at the first wave rather than "better
than average" over time by 19.8 percent (OR =. 802, p<.01). A higher education
enables the preservation of better-than-average self-rated health because it enables
one to acquire, understand, and act on information to that end.

Higher income contributed to better maintenance of self-rated health over time.
Having a higher income level reduced the odds of people rating their health worse
than average at both waves rather than rating their health better than average at both

waves by 10.8 percent (OR = .892, p<.001). Having a higher income contributed to

120

provide financial security to the elderly; this in turn positively influenced their self-
rated health. Thus, although the effects of socioeconomic variables were weaker than
a multiple roles variable, they contributed to the maintenance of better self-rated
health over time.

Age influenced self-rated health in the opposite direction compared to a multiple
roles variable and education. Table 9 shows that additional age measured in years,
increased the odds of people rating their health worse than average at both waves
rather than better than average at both waves by 4.1 percent (OR = 1.041 , p<.001). It
also increased the odds of people rating their health better at the first wave rather
than better than average at both waves by 4.3 percent (OR = 1.043, p<.001). These
results suggest that becoming older increased the likelihood of individuals
downgrading their self-rated health over time as well as increasing the likelihood of
people continuing to rate their self-rated health worse than average over time.
However, Table 9 also shows that additional age increased the odds of people rating
their health better at the second wave rather than better than average at both waves
by 2.1% (OR = 1.021, p<.001). These results suggest a perception of recovery is
possible but less frequent than the perception of worsening health.

Finally, marital duration slightly, but negatively, influenced self-rated health over
time. Table 9 shows that longer marital duration measured in months slightly
increased the odds of people rating their health average at the time of both waves
rather than rating their health better than average at the time of both waves (OR =

1.002, p<.001). Longer marital duration also slightly increased the odds of people

121

rating their health worse than average at both waves rather than rating their health
better than average at both waves (OR = 1.002, p<.001). Longer marital duration
also slightly increased the odds of people rating their health better at the second
wave rather than better than average at both waves (OR = 1.001, p<.05). These
results imply that marital quality has complex effects on health rather than a simple
accumulation of effect with time. Since this data set does not include variables
measuring marital quality, future studies are necessary. It is possible that longer
marriage does not increase marital quality or happiness, but rather increases boredom
and discontent (U mberson et al. 2005).

o 3 — C: Psychological Health (Table 10)

Table 10 summarizes the results of the multinomial logistic regression, examining
changes in the depression scale between the waves. Consistent with the salubrious
effect of having many diverse roles, I find that those who had more roles were 15.4
percent less likely (OR = .846, p<.01) than those with fewer roles to remain
depressed between the two waves than to remain free of depression. In addition,
respondents who lived in a household with more residents had much lower chances
of experiencing the onset of depressive symptomatology between the waves or of
maintaining depressive symptomatology rather than remaining free of depressive
symptoms (OR = .905 and .863, p<.001, respectively). These results are consistent
with the result of my cross-sectional analysis, which showed that multiple roles
reduced the risk of having more symptoms of depression, controlling for the effects

of other variables. These results answer Research Question 3 affirmatively.

122

The control variable also mattered. A higher income level reduced the odds of
continuing to have some symptoms of depression rather than continuing to have no
symptoms of depression by 5.7 percent (OR = .943, p<.05). This result suggests that,
although the positive effect on mental health is weaker than a multiple roles variable,
financial security is important in maintaining better mental health over time for the
elderly in Japan. A higher income level also enables the elderly to spend more
money on leisure and transportations.

Age influenced the mental health of the elderly in the opposite direction
compared with a multiple roles variable and income. Table 10 shows that,
controlling for other variables, additional age measured in years increased the odds
of continuing to have some symptoms of depression rather than continuing to have
no symptoms of depression by 2.1 percent (OR = 1.021, p<.05). It also increased the
odds of an onset of depression rather than having no symptoms of depression over
time by 3 percent (OR = 1.030, p<.01). This result suggests that the risk of
depression rises with age, perhaps in response to increasing disability. Alternatively,
Umberson et al. (2005) found that older respondents are less likely to be emotionally
resilient because more experiences gained over the life course often lead to less
conflicts in human relationships.

Finally, Table 10 indicates that controlling for other variables, residence and
gender influenced the psychological well-being of the elderly over time. Controlling
for the effect of other variables, compared with men, women were less likely to

continue having any symptoms of depression rather than continuing without

123

symptoms of depression by 36.8 percent (OR = .632, p<001). Controlling for the
effects of other variables, compared with men, women were less likely to report an
increase in symptoms of depression over time rather than to maintain no symptoms
of depression by 34.2 percent (OR = .658, p<.01). This result suggests that the
gendered life course in late life may benefit women more than men in Japan. As
previous studies suggested, women are better at maintaining and extending social
networks than men in the late life course (Antonucci and Akiyama 1995) and women
are more likely to be socially protected than men in the later life course (Ikeda et al.
2007). This finding also refutes Durkheim's gendered assumption that women are
less likely to obtain either the benefits or the costs of that involvement, especially
considering the later life course of the elderly in Japan.

Residence showed inverse effects on psychological health compared with
physical health. This result was consistent with my cross-sectional analyses of
multiple roles and health (Chapter 5). Compared with rural residents, urban residents
were 37.6 percent more likely to continue having some symptoms of depression
rather than continuing to have no symptoms of depression (OR = 1.376, p<.01).
Again, this result supports those of Fukuda et al. (2005), who found that urban
residence provided a less supportive social network than rural residence in Japan.
This result suggests that not only is occupying diverse roles important for the well-
being of the elderly over time, but also being in a context where neighbors can
support the elderly is important. In 2003, 19.3 percent of the elderly lived alone in

major cities compared to 10.3 percent in rural areas (Ministry of Health Labour and

124

Welfare 2003). The number of solitary deaths (kodokushi) increases as the number of
the elderly living alone increases. Although there is a system of care for the elderly,
ten years after the Great Hansin Earthquake that hit Kobe city in 1995, the number of
kodokushi totaled 560, of which 32 were suicides; 11 of these were discovered over a
month later, and one was found after a year (Shiozaki 2005). Thus, supportive
communities and neighborhoods become more important for the elderly to avoid

social isolation, especially in urban areas.

Summapy

To summarize, longitudinal analyses of the first wave and the second wave of
NUJLSOA, the results presented in this chapter support the role enhancement
perspective. The answer to the third research question is that the elderly who had
diverse roles at the time of the first interview were more likely to remain healthy at
the time of the second interview. This statement was supported for all health
measures including physical health, self-rated health, and psychological health
outcomes. As previous studies of the role enhancement perspective emphasize
(Wethington et al. 2000), occupying multiple roles leads to higher levels of physical
and emotional health than do fewer roles. These results also support Traphagan's
claim that, to be a good elderly person in Japan, one needs to be a socially integrated
individual who engages in activities that involve social interaction. Failure to
maintain social involvement invites a deterioration of health in late life (Traphagan

2004).

125

In addition, supporting the life course perspective, a prior positive experience,
measured by a multiple roles variable at the first wave, positively influenced the life
trajectory of health at the time of the second wave. The elderly with diverse roles
were less likely to experience transitions to better or worse health conditions at the
time of the second wave than were the elderly with fewer roles because they were
more likely to remain healthy over time. As Moen et al. (1992) claim, successful
aging is a cumulative process, and social integration during the adult years appears to
promote both social integration and health in the later life course.

The effects of age, gender, and residence on changes in health between waves
remained, after controlling for other variables. However, how these three variables
influenced the well-being of the elderly was by far more complicated than previously
reported. In particular, it is interesting to notice that, compared to rural residents,
urban residents had a lower risk of having some physical disability at the time of the
first wave (Chapter 5) or of becoming disabled by the time of the second wave
(Table 8). Urban residence increased the risk of having depressive symptoms at the
first wave rather than no symptoms (Chapter 5) and of remaining depressed rather
than undepressed across the waves (Table 10). Residence, however, did not influence
changes in self-rated health (Table 9). The disadvantage faced by rural residents may
reflect the selective effects of mortality. That is, rural Japanese survive to older ages
than urban Japanese and thus are exposed for a longer time to the risk of living with

nonfatal but disabling conditions (e. g., arthritis). The advantage enjoyed by rural

126

Japanese of avoiding depression longer than urban Japanese may stem from what I
argue is a cause of their greater longevity.

Gender influenced changes in psychological health, but not physical disability
and self-rated health (Table 10). In the United States, women have worse self-rated
health than men but are less likely than men to die at each age (Case and Paxon
2005). In Japan, the results showed that elderly women were less depressed than
elderly men were, and perhaps for this reason, on average, Japanese women live
longer than Japanese men do. My study suggests that the life course in late life is
gendered and benefits women more than men in Japan.

The average life span of Japanese women grew .29 years from 2005 levels and
became 85.81 years, maintaining the top position in the world for 22 years
consecutively (Kyodo News 2007). Although Japanese men do not live the longest in
the world, they lived 79 years on average in 2006, growing .44 years from 2005
levels and rising back to the second position from the fourth position in 2007 (Kyodo
News 2007). Faster growth of the life expectancy at birth for men than women and a
shrinking gap in life expectancy between men and women could be a reason for
insignificant differences in self-rated health and physical disability by gender.
Continuous efforts to maintain the mental health of female elderly in Japan may be a
reason why Japanese women live the longest in the world and still live longer than
Japanese men do.

The cross-sectional findings (Chapter 5) and longitudinal results (Chapter 6) not

only provide policy implications to promote healthy aging, but also show limitations

127

in the current data set that are answerable in future studies with future waves or a

new data set with more variables. These will be discussed in the next chapter.

128

CHAPTER 7
CONCLUSION

Aging in Japan in the Twenty-First Century

This chapter summarizes my dissertation, discusses the policy implications of the
findings of this study, and calls for future research on aging based on the findings
and limitations of this research. First, I briefly summarize historical changes in the
understanding of aging and families in Japan, and then discuss how my findings
from cross-sectional and longitudinal analyses of the Nihon University Japanese
Longitudinal Study of Aging (NUJLSOA) to expand knowledge of aging in Japan,
and how our understanding will transform in the twenty-first century. Then, based on
my findings, I discuss implications, including government policies to promote
healthy aging. Finally, I will discuss how future studies can extend the findings and

overcome the limitations from this study.

Summagy of My Dissertation

Scholars have discussed aging in Japan as a result of the demographic, cultural,
and economic transformations after World War II that changed the balance of
intergenerational caregiving (Asai and Kameoka 2005; Maeda 2004; Ogawa and
Retherford 1993; Retherford et al. 1996). The major problem in some of this

literature is the historical starting point scholars choose.

129

lntergenerational mutual care in Japan developed in the context of various factors,
including cultural ideals, centralization of the civil state, and the family unit called
the ie. Prior to the Meiji period, there were regional and class diversity in family life
and no cultural ideal for women to take care of the dependent elderly. In the
Tokugawa period, care was often emphasized as men's morality in the public sphere
because the Tokugawa shogunate emphasized the Confucian ideology of filial piety.
As Japan moved from the Tokugawa to the Meiji period, it became more feasible for
the national government to create legal pressures on women to stay at home to care
for the children and the dependent elderly in the privatized ie.

Many scholars start with this patriarchal ie system (the system of primogeniture
legally recognized in the Meiji Civil Code in 1898) as the benchmark against which
to gauge the continuity, uniqueness, and change of eldercare during the twentieth
century (Asai and Kameoka 2005; Long and Harris 2000). This resulted in many
scholars assuming that the cultural underpinnings of the ie defined the care of ‘fiail
elderly’ as ‘women’s work’ undertaken to preserve ‘women’s morality’. In such a
context, the discussion of elder care is heavily focused on the burdens of women,
explicitly or implicitly categorizing care receivers as passive recipients of care. Since
care has been pessimistically perceived as a caregiver’s responsibility and obligation
for looking after someone weak and dependent, the elderly have too often been
associated with burdens and problems, and stereotyped as sickly, frail, and

dependent.

130

Owing to longevity and reduced disability among the elderly after the Second
World War, more elderly challenged the restricted and pessimistic definition of care
from middle-aged women to dependent and frail elderly. The elderly are seen
increasingly as proactively choosing their opportunities to care and be cared for by
important people beyond the family in their later life course. Such social recognition
has broadened studies of aging beyond just the maintenance of physical and mental
vitality to include the social integration of older people with family, friends, and
communities. In this context, research on social integration and health has emerged
as an urgent and important issue of scientific as well as of public concern for
promoting healthy aging in Japan, as well as in other countries of the world.

0 The Influence of Social Integration on Health

The most traditional way of theoretically and empirically explaining the
relationship between social integration and the well-being of the elderly is to take
role involvement as an indicator of social integration. Two different explanations are
the role strain perspective and the role enhancement perspective. Previous studies
found evidence that the role enhancement perspective is more applicable to older
adults than the role strain perspective. For the elderly who have been through various
life events such as marriage, childrearing, and deaths of their parents, holding diverse
roles benefits their well-being by increasing individual resources, social connections,
and social identity, as well as by providing more “chairs” that the elderly can fall
back on. The benefits exceed the costs to their well-being arising from more role

conflict or role overload that comes with more diverse roles. Incorporating the life

131

course framework that emphasizes the importance of prior experiences in
determining later health outcomes, plus the importance of various contexts
influencing the late life course, I developed three research questions to study the
relationship between social integration and health of the elderly in Japan.

Both cross-sectional and longitudinal analyses of the two waves of the NUJLSOA
supported the role enhancement perspective, not the role strain perspective. The
elderly who occupied diverse roles had a greater likelihood to be healthy than those
who occupied few roles at the first interview (Research Question 1). This result was
consistent for all three health measures: physical disability; psychological health; and
self-rated health. Holding diverse roles benefited the well-being of the elderly,
controlling for age, gender, residence and socioeconomic variables. For the elderly
who went through various life events such as marriage, childrearing, and deaths of
their parents, rather than increasing role conflict or role overload, occupying diverse
roles was more likely to benefit their well-being by enhancing individual resources,
social connections, and social identity, as well as by providing more “chairs” that the
elderly can fall back on.

Looking at the individual role and role density in the first wave, I found that
certain roles had a significant influence on health (Research Question 2). Although a
diversity of roles contributed to the elderly’s feelings of value and security, “density”
in a given role — shown by having more sons, daughters, stepchildren (parental role),
and grandchildren (grandparental role) — did not positively influence the well-being

of the elderly. Perhaps the reason is that this data set is missing information about the

132

quality of the relationships for each of these four categories of relations. However, a
greater participation in social activities and groups showed a significant influence on
the well-being of the elderly in the first wave. This result implies that Japanese
elderly who proactively choose their opportunities to care and be cared for by
important people beyond the family remain healthier than those with fewer
extrafamilial memberships. This result is also consistent with the findings of a US.
study which showed that being a member of a social organization appeared to be
especially salutary (Moen et al. 1992).

Supporting the life course perspective, prior positive experience measured by a
multiple roles variable at the first wave positively influenced the life trajectory of
health at the second wave. The elderly who occupied diverse roles at the first time of
the interview had a greater likelihood to remain healthy at the second time of the
interview (Research Question 3). The elderly with diverse roles were less likely to
experience transitions to worse health outcomes because the effect of multiple roles
on health lingered, and they were more likely to remain healthy over time. This
result suggests that successful aging is a cumulative process, and that the life course
fi'amework is a useful frame to capture the relationship between social integration
earlier in life and health in later years.

0 The Influence of Residence on Health

Residence showed complex effects on health outcomes. The rural elderly were

more likely to have a disability than were the urban elderly. However, the rural

elderly were less likely to feel depressed than the urban elderly. Since self-rated

133

health takes into account the opposing effects of physical and psychological well-
being of the elderly, residence did not show a significant net effect on self-rated
health. These findings suggest that a greater homogeneity in rural areas benefited the
mental health of the rural elderly, whereas unequal access to health and social
services due to depopulation in rural areas disadvantaged the physical well-being of
the rural elderly. Alternatively, perhaps the longer survival of the rural elderly
exposes them to a higher risk of developing a nonfatal but disabling disease that
results in physical disability (e.g., arthritis) than it does for the urban elderly. Future
research including more consecutive waves of the NUJLSOA should explore how
residence (rural/urban) and health statuses at Wave 1 are related to the progression
into worse health or death at later waves. My dissertation has demonstrated the
usefulness of multinomial logistic regressions in this exploration.

Such complex results of rural and urban differentials on health outcomes have
also been found in US. studies. As in Japan, rural Americans have a higher rate of
disability than their urban counterparts (Hummer et al. 2004), but they die at older
ages than their urban counterparts (McLaughlin et al. 2007). This may be due to
selective outrnigration of rural Americans to urban places where they die more
quickly than the rural people they leave behind. Future research in Japan and the
US. is needed to find whether selective outrnigration is the reason why the rural

elderly in both nations appear to live longer than their urban counterparts.

134

o The Influence of Gender on Health

Elderly women were less likely to be depressed and more likely to remain
mentally healthier than their male peers. These results suggest that the later life
course benefits women more than men in Japan, although women may be more
likely to suffer from role conflicts until they finish raising children and taking care of
parents. As previous studies suggest, women are better at maintaining and extending
social networks among and beyond family members, which could be a reason why
Japanese women live longer than do Japanese men. Networks women have built in
their family and in their community over their life course have a positive effect in the
later life course relative to the networks men have built in their work place because
the latter social network is more likely to disappear after their retirement.

These findings indicate the importance of government policies to provide
opportunities for the elderly, especially elderly men, to extend and to remain in a
circle of harmonious integration (wa) where people can care for each other beyond

the family, particularly in urban areas.

Poligy Implications
o Discovering ikigai ("purpose of life")

Analyses of the first wave of the NUJLSOA showed that greater participation in
social activities and social groups had a significant relationship with all the health
outcomes. Both cross-sectional and longitudinal analyses of the data suggest that the

elderly who occupy more diverse roles remain healthier than those who are more

135

socially isolated. Thus, the first policy implication from the findings of this study is

to improve and increase programs, facilities, and policies to enhance the ikigai
(i ébilfi) of the elderly.

For the elderly to discover their ikigai, there are many unique facilities in Japan.
For example, according to Senuma (2006), there are 50,000 courses for the elderly
taught at universities and schools. If 100 people, on average, participate in one
course, there will be 5,000,000 elderly learning at school or through the media. A
long history of high literacy rates across social classes beginning in the Edo period
may have influenced the popularity of these facilities for the elderly. However, the
problem or the challenge Japanese society faces is that, although there are many
elderly willing to take courses, the number of these facilities has not caught up with
the increasing number of the elderly. Since many colleges and universities in Japan
will continue to lose students due to low fertility rates since the Second World War,
developing life-long education programs at colleges and universities is an effective
way to increase students and diversity at colleges and universities in Japan.
Developing life-long education programs at colleges and universities will also
increase opportunities for extending intergenerational networks between the elderly
and young adults. In addition, the government should develop programs to frmd
elderly students because many of them could not obtain higher education due to
wars. Thus, not only increasing educational facilities and programs for the elderly

but also funding their tuition will boost their enrollment. To promote healthy aging,

136

such governmental funding to support the elderly is as necessary as developing day
care services, rehabilitation facilities, and nursing homes for the elderly.
0 Revitalizing rural villages

Depopulation became a major problem for many small villages in Japan as
fertility rates have declined in the postwar era (Knight 2003; Thompson 2003). The
deficiency of funds to provide needed health and social services is a symptom of the
negative long-term effects of depopulation (Thompson 2003).

Recently, some rural municipalities have found a way to repopulate. Across
Japan, some rural municipalities have established assisted living facilities that
include special housing, customized medical facilities, and home helpers for older
adults(Knight 2003). In developing these facilities in rural areas, both the national
government and the rural municipality must listen to the voices of local people, who
may fear that the rural village will become a dumping ground for the elderly. While
popular in the United States, assisted living centers are new to Japan and may be
resisted by Japanese people as same as nursing homes or hospitals. Thus, in
developing assisted living facilities in rural areas, local people need to be involved in
order to understand benefits of welcoming these elderly in rural areas. The elderly
moving to rural areas can be great contributors to enhance social integration in rural
areas by increasing new elderly friends for the rural elderly and by increasing
budgets to develop educational, recreational, and medical facilities for the elderly in

rural areas. At the same time, it is important to improve security systems for the rural

137

elderly living alone as well as for local municipalities to support the grass-roots
movements to repopulate the village.
0 Finding a way to reduce kodokushi (solitary deaths)

The number of kodokushi (BIKER) increases as the number of the elderly living

alone increases, especially in urban areas. Issues regarding kodokushi became a vivid
problem when the Great Hansin Earthquake hit Kobe city in Hyogo on January 17,
1995 (Shinozaki, Nishikawa and Deguchi 2005; Shiozaki 2005). Although there are
no comprehensive data showing the number of kodokushi, Professor Ueno, who
totaled the number based on newspaper accounts, estimated that over 200 kodokushi
were lost in Hyogo (Shinozaki et al. 2005). Men were twice as likely as women to
die of kodokushi, and middle-aged males (55-64 years old) constituted over 40
percent of those who died alone (Shinozaki et al. 2005). Such tragedy taught
everyone that residential areas could be either supportive or risky environments for
older adults. Japan and other nations have the possibility of encountering
earthquakes, and all nations in the world have the possibility of natural disaster such
as storms and heat waves.

In fact, a great heat wave struck in July 1995 in the city of Chicago; and people,
especially people living alone, died solitary deaths (Klinenberg 2002). The ethnic
and racial differences in mortality also taught everyone about urban life. The actual
death numbers for African Americans and whites were almost identical. However,
when the differences in size and age composition of African Americans and Whites

were taken into account, the Chicago Public Health Department found that the

138

black/white mortality ratio was 1.5 to l (Klinenberg 2002). Latinos, representing
about 25 percent of the city p0pulation, accounted for only two percent of the heat-
related deaths (Klinenberg 2002).

Natural disasters such as the Great Hansin Earth Quake and heat wave can hit any
place in the world, and people who are vulnerable to these problems are often the
elderly, especially the elderly living alone in urban areas. Findings from my
dissertation research suggest that, not only in rural areas, but also in urban areas,
building a better community, which enables the maintenance or extension of social
networks beyond the family, provides important grounds to promote healthy aging.
Thus, it is necessary for national governments and local municipalities to support the
grass-roots movements to build a better community to prepare for natural disasters.
Developing programs and building facilities where people from the community can
gather and chat, cleaning the community together to avoid crimes and to improve
neighborhood networks, and improving the security system of houses for the elderly
living alone should be developed by national governments, local municipalities, and

local people before it is too late.

Implications for Future Research
Since only the first two waves of the NUJLSOA survey (1999 and 2001) were

available at this time, there were too few cases of mortality for me to analyze the

relationship of death to social integration, and to examine how this relationship

139

differs by age, gender, and residence. There were also too few cases to examine
whether changes in the number of roles or the loss of role diversity might influence
the relationship between social integration and health. Further waves in this data set
will enable me to test hypotheses regarding social integration and mortality. Since
the elderly in both waves are relatively healthy on average, the passage of more time
since Wave 1 will yield more deaths and may reveal whether having a larger number
of sons or daughters enhances longevity beyond the fact of being a parent. To
understand how the quality of relationships influences the well-being of the elderly
over the life course, qualitative studies are also necessary to understand differences
in the relationships the elderly have with their sons and daughters as well as with
their marital partners.

Further comparative studies looking at morbidity and mortality between the urban
and rural areas are necessary to examine whether the urban elderly live longer than
rural elderly, despite the findings from this study that the urban elderly are more
likely to be disability-free. If headaches, arthritis, and vision problems are much
worse for rural than urban elderly, the rural elderly could plausibly be more likely to
report themselves as disabled. Future studies should look at specific health problems
to understand nrral-urban differentials in aging. Further comparative studies looking
at morbidity and mortality between men and women are also necessary to understand
how the effects of gender change as elderly respondents become older. Although my
dissertation research could not find any significant effects of gender on physical

disability or self-rated health of the elderly, further waves available in the future may

140

discover significant gender differences in the relationship between social integration
and those dimensions of health.

I used multinomial logistic regressions to analyze the data set, and this statistical
approach limited the number of categorical variables that I could use as predictors of
health measures. A promising predictor of social integration being examined by
scholars at Nihon University is living arrangements. These studies will show whether
elderly people in three-generation households are healthier and more integrated than
those in two-generation households.

I also realized possible cultural limitations in using IADL items to assess physical
health of the Japanese elderly because some of the items (e. g. preparing meals,
taking care of financial matters, and dusting and cleaning the house) could be gender
specific. Looking at descriptive statistics, 1 saw no noticeable differences between
the elderly men and the elderly women in answering “Do not perform” to the
question whether they take care of financial matters (5.9 percent for men vs. 4.4
percent for women) and to the question whether they dust, clean up, and do other
light housework (3.0 percent for men vs. 1.2 percent for women). However, there
was a noticeable difference between elderly men and elderly women in answering
“Do not perform” to the question whether they prepare own meals (20.7 percent for
men vs. 6.1 percent for women) although the majority of both men and women
answered that they have no problems in preparing their own meals (72.1 percent for
men and 85.3 percent for women). Future studies using further waves should

examine the cultural appropriateness of using measures of disability developed in

141

Western nations and should pay special attention to gender differences in answering
IADL items.

Theoretically, my dissertation research supported the role enhancement
perspective rather than the role strain perspective. My dissertation research also
found that the later life course benefited women more than men, and it refuted
Durkheim’s assumption that women are less involved in public life than men and
thus less likely to obtain either the benefits or the costs of that involvement.
Women’s higher educational attainment and longer involvement in the labor force in
contemporary Japan than in 19th century Europe may have increased the benefits and
lowered the costs of role involvement. Thus, to theoretically advance the study of
social integration and health, future examinations of the NUJLSOA should seek
whether women’s educational attainment and duration of formal employment
increase across the waves and are associated with social-role diversification and
improving health. In addition, as discussed in the theoretical framework, previous
studies have looked at the role enhancement perspective and the role strain
perspective as two competing perspectives. However, it is possible that occupying
diverse roles may become a positive or negative factor depending on different
phrases of the life course. To theoretically advance the study of social integration
and health, it is necessary to incorporate the life course framework so that future
studies can examine whether the role strain perspective or the role enhancement is

supported in different phrase of the life course.

142

Future studies should also involve data sets of other nations to understand the
relationship between social integration and health in the later life course and to
examine how this relationship changes with age, gender, and residence. Qualitative
studies of the elderly in vulnerable circumstances - men, older, living alone, low
education level and income, and poor social networks - are necessary to make them
visible to the society and bring their voice to public. Enriching the study of social
integration and health with further waves, as well as with data sets from other
nations, enable us to globally promote healthy aging in the twenty-first century.

To conclude, the culture of care was historically diverse, neither strictly nor
naturally assigned to women in the private sphere. The culture of care became
private and “women’s work” after the Meiji Period; however, it has reentered the
public sphere owing to the longevity and reduced disability among the elderly. The
responder to the needs of the elderly should consider both the care receiver’s and
caregiver’s viewpoints. In discussing the culture of care in the future, it is important
to avoid stereotypical assumptions that the elderly are frail and dependent.
lntergenerational mutual care in Japan developed as a result of the demographic,
cultural, and economic transformations, and it will continue to transform in response
to cultural, demographic, political changes. In thinking of intergenerational mutual
care in the future, instead of focusing on drawing boundaries between men and
women, public and private, independence and dependence, scholars and policy

makers must realize and accept a fluid situation and to understand social integration

143

embedded in period-specific cultural, political, economical, and geographical

contexts.

144

APPENDICES

145

Figure 1

Map of Japan

North East

 
   
 

Central

:4

146

Figure 2

Population of Japan

 

(1.000 people) Population ofJapan

 

140000

:: A
.... I \
6.. J \

 

 

 

 

 

 

 

 

 

20000 K/
0 IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
to .5:

«8‘06 '01:?9‘19553 .I

 

(Years)

 

 

Source: Kito, Hiroshi. Jinko Kara Yomu Nihon No Rekishi (Demographic Analysis
of Japanese History). 7 ed. Tokyo: Kodan-sha, 2001.

147

Figure 3

Total Fertility Rate (T FR), 1925 to 2004

 

 

TFR (1925 to 2004)

 

 

 

 

lllllllllIllllllllllllllllllllLllllllJlllllllIllllijlllllllllllll

W'TTIII'IUT‘IIUIITITIIIIIFIII—IIIIIllllIUIUIIIIIIIIUIUIIIIIIUITUI]

 

 

 

 

 

 

 

0

1925 1940 1949 1954 1959 1964 1969 1974 1979 1984 1989 1994 1999 2004

 

 

Source: National Institute of Population and Social Security Research 2006

148

Figure 4

Labor Force Participation Rate by Sex

 

 

 

 

 

 

 

o I I I l l l l I l l l l 1 l l I l l 1 1 J l o
umxwfiwfiwfiwfim fiNfiMfiflfiWfiwfiW
llllllllll+ lllllllllll+
”anywuexwaw wuwynuwfiwam

Source: Figure 12.1 taken from Chapter 12 Labor in Statistical Handbook of
Japan, Statistics Bureau & Statistical Research and Training Institute, Ministry of
International Affairs and Communications. Available on-line
[http://www.stat.go.jp/

english/data/handbook/pdf/cl2cont.pdf] (accessed on November 1, 2006)

149

Figure 5

Life Expectancy at Birth for Japanese Men and Women

 

LEB (yrs)
90

 

 

 

80

 

 

70

60

 

 

 

50

 

 

 

J L 4 A l 1 I l L I I A l I I I l L L
I I I I I I I I I I I I I I I I

 

1921"25 1947 1960 1975 1990 9001 9095

 

 

 

Source: National Institute of Population and Social Security Research 2006

150

Table l
Paired-Sample t-tests of dependent variables

 

Time 1 Time 2

 

Variable M @D) M (81)) d) t Sig._

 

Self-Rated Health 2.8358 1.10 2.8820 1.09 4004 -2.611 .009

 

Psychological Health 1.2772 2.07 1.3966 2.32 2869 -2.587 .010
(CBS-D Scale')

 

 

 

 

Physical Health .7244 2.47 1.0394 3.02 2948 -8.360 .000
(ADL+IADL)

ADL2 .3181 1.19 .4728 1.46 3908 -8.249 .000
IADL3 .4197 1.38 .5861 1.65 2991 -8.075 .000

 

 

 

 

 

 

 

 

 

 

2

3

Depression measures (CBS-D) include nine items: 1 Didn’t have much appetite, 2 Felt depressed, 3
Ordinary things felt troublesome, 4 Trouble sleeping, 5 Felt lonely, 6 People seem unfiiendly, 7 Felt
sad, 8 Felt hated by others, 9 Trouble feeling motivated. Two points was given to anyone who
OFTEN found these items to be true during the past week. A point was given to anyone who
SOMETIMES found these items to be true during the past week. No point was given to anyone who
RARELY felt these items to be true in the past week. The score ranged from a theoretical low "0" to
a theoretical high of "14."

ADL measures include seven items: 1 Bathing or showering problems, 2 Dressing, 3 Eating, 4
Standing up fi'om a bed or chair/sitting down on a chair, 5 Walking (around the house), 6 Going
outside (leaving the house), 7 Toileting. A point was given to anyone who answered who had
trouble with one of these items. The possible range is "0" (no difficulty performing IADL items)
and "7" (difficulty in performing ADL items).

IADL measures include seven items: 1 Preparing meals, 2 Leaving home to purchase necessary
items or medication, 3 Taking care of financial matters (paying utility or newspaper bills), 4 Using
the telephone, 5 Dusting, cleaning, and other light housework, 6 Taking the bus or train to leave
home, 7 Taking medication as prescribed. A point was given to anyone who answered who had
trouble with one of these items. The possible range is "0" (no difficulty performing IADL items)
and "1" (difficulty in performing IALD items).

151

Hugo N” 7900:0800 roman.“ Wofiommmoa Ooaumibm 20 ”30000008 $0.00 006 9. 82.6 00m00=§ 0: EH 9.03:. 838. 90
Wamvoaoam $09 0» ~68» one c». 90 002,09. 0: Bub 0:0 SUE mom? ammgm. «50 00 Womwoaaam $00.. no 030080 0095me

 

 

 

936808 Gamma—.5
One 0.. 50.6 00.3an 20 2.2.8..
102.0020 Ow mm E. 0! mm 0..
u 5 nm . _ — _ _.
3:50.» ”0.8 .0001... .000 _ .000: .000 _
0 0
65.000: 0 _
3018. 0:300: #001... .000 _ _.000 .000 _
008— 0:3000 o», 2255 .000 .000 _ _.00~ .00.). _
macaw—nae:
ra<o_ 0». 8:82 €00. 00:08: .000 .30 0 .000 .50 _
Faun 09.3.5032:
2:802. cm 320003 =<m=m Homage. 3 9o #03: .0N0 _ r303... .80 _
meagre—0
E
N00800: .30.... .000 _ .302. .000 _
58:5 .000 .0N0 _ .00 _ .08 _
$0» 2 ~01... .000 _ _ .0001... .000 _
”9.0033 3.020 .03.... .000 0 .000 .000 _
gown—.008” ”=3: . . 0 . . 0
wax 9030—00 #30 .20 _ 0.00.81. ._0_ _
960.0808” 205 0 0

 

 

5.8.0500 300 05-358 0 000.000“ 006.9% om 0.8005 n N0“ 0 n 300 A<m=0 0880.
208% O” n 000m 809 mmnm8=0m~0 and... .38 Homo—6:8 same—.0. mm «802.038 $00. :0 03008— 3500008:

nflAbM 1A0: nan—Xbo—

anagoa 2.0 max 2.0 88mo18. 3000088050 500030o3 £5028. 0:0 3. 05 $.53 a no" 8 N63 080:8 0 mm 805093.

152

0.00.0 0. 30.000800. 50.000 ”000000.00 008000.00 3. ”0000000000 200 00.00 0.0:. 00000. 0000.. 0:0: 0.0.000. E
”0000:0030 $0.0 00.00 90... 0005. $0000 000: 00.000. 000 A00 W0000000:.0 200 00.00 0.0.0 .00.... 0.00.000 9000.00.00

 

 

 

0080on
.0032. 0.0: 0<0..0m0 $2.00 0.0: 0<0..0m0
«00300.0 0% mm 00 ON mm 00
u 5 nm . _ n _ _.
30.00... ”0.00 7.00.... .000 . .0001... .000 .
0 0
55.000: . .
300:0. 9.0000: .000... e .000 . . .00. n .000 .
0.08. 000000: 0.. $000.5 .000 .08 . .000 .08 .
90.000020:
.05. 0.. 00:88 50.... 0.0.0.0: .00. .. .0 . .000 .. .. .
5.0:“ >30:m0.:0:.
2.5.000 0.. 800.0000 ..<.:m 0800.00 .: 0.0 .000 .08 _ .08 .08 .
00:00:20
EBB
m000000: . .00. .000 . .03 .00 . .
502:0 . .08.. .0. 0 . .000... .80 .
>00 .000... _._ .000 . . .0. 0 ... .000 .
”000300 20.00:. .000 .000 . .000 .000 .
9000.030. 5.00.0 . . 0 . . 0
m0x 20030.0. :00 .000 . ...0_ .000 .
0 0

900002.00. 2.0.0.

 

 

5.8.0.000 .000 0.0000000 0 000.0000 00m0000 00 @0000... u N0. : .l. 0000 A<0..0 00000..
20.00” Ow n 0000 0000. wmu0§00~0 2.3... .....0 00.000000 00.00000. .0 0000000030 £00 .080 0.0... 000.9 0<0..0m0

n0A.00. 1&0.“ 0.00A00.

w00.00:00 0:0 m9. 000 00.000100. 3.0.5.0505... E0000:00:. 0100.00. 0:0 0.. 0.0 0.08 .0 00» 8 N03 0000000 0 .0 000.500....

153

0.00.0 0. 700.035.... €0.30 ”00800.0: 00800050 3. ”0000:0030 2.0. 0:0 0.. 80.0 000.00 0: 00000008: 000.0. 0:0 3.

”0000:0020 2.0. 0. .000. 0:0 08 80.0 8.00.:m 000.00 0: 00000008: 000.0. 0:0 3. ”0000:0030 8.... 00.0 000.0 0: 000.0008:

 

 

 

 

000.0 900.0830 00.0000...
<0...0...0 0:0 0.. 50.6 000.00 0: 00380.0: 20 >512.
O” m... 0n 6” mm 00
5.000... 00 @000. .:. .000:
3500.0 ”0.00 .0000... .000 . .0003... .000 .
0 0
9.0000: . .
2.0....0. 05.000: .00.... .000 . ..00. .000 .
0.0.0. 05.000: 0.. £00w.:m .000... .08 . .0001. .00... .
90000000:
.x0<0. 0.. 00:80. 2.0. 0....000: .000 .00 . .0001... .00 .
500:0 2.030033.
25.00.. 0.. 50:50.0 ..<.:m 8000.00 .: 0.0 .0003... .0~. . .000 .000 .
30000020
00:20. «.010—000
0.000000: . .000 .03 . .000 0 .00. .
50030 .00 . 0...... .0 . m . .000 .08 .
>00 .03 .000 . .03... 2.. .00.. .
”00.0300 330:. 02.0.5... .0... . .000 .000 .
90000300. 5.00.. . . 0 . . 0
wax 00080.0. .00. .2 .000 . .bmm. ..0. .
0 0

9000.030. 7.0.0.

 

 

5.8.0.000 0000 00.000000 u 00... .00. 000.000 0.. @0000... u no. : n 00.3 A<0..0 00000..

20.00. 0” n 0000 0000. mmn0.0:00~0 0.400. .....0 00000030 00.0m0Q .0 8000:0030 :00. ~03 00000 0: 00000008: 000.0

00A.0u. 15.0.. 2......A.00.

”00.0300 0:0 m8. 000 00.000000. 3.0—5.0.0000. 90003032010200. 0:0 0.. 0.0 <0...00 .0 00. .0 0000 0000000 .. .0 000.500....

154

Hus—a V? 3:583». rommmmo womqammmoa 2.09% 003315 CV womvoaaaam SE. 95 3 :58 385.3% 0: Eur 9. 59.. .833 35

A8 womvosqosfi SE. so ciao”: “=37sz gown—6:8 0&3on

 

 

 

 

32.n- — 32.2 N 323. u 32.». a.
5538 OW mm a». 0% mm a». 0% mm a». 0% mm mm
m ._ z _ _ .
39:8— mSEm” 22 32.13 .mce .Ea _ .wma .fiw _ .mm@ .33 _ .mqa .Su _
93,238” 32453 c o o o
338— 0:839. focc boo _ b3 .25 _ race boo _ . Eco boo _
2532, om 85m Zea .oqw _ roeu .3.» ~ :3 .3» _ foes .cd _
2.532 cm gnaw—.83 :3... .od _ :3... bum ~ 58.. .03 _ :3... baa _
2553.. cm mat 9:96: bum .omN _ bum .omw ~ .3» .omw _ .03 .omu _
mmammwnaos s: 0358: bout... .73 _ 25...: .T8 _ .39.: .73 _ 08...: .3.» _
2:302. om gang—9o: bow .oma _ boo .omq H .02 .Sa _ .emc .Sq _
2:53 a. 303%; 55m 88 Sm _ .23 .80 _ 28 .8“ _ .93 .80 _
"£092 5 :8 3050305
__._ _. :— _ _.
€o~r mSEmu 29 fienabm 93%.: .3_ _ “WE—I... Law _ 98%.... .3— _ 93%;... .Ew _
93238“ $029.8 . . c . . o . . o . . o
.38— 95303 o». £0135 :5— .oow _ #25 be.» _ r2: bow _ .bco bow _
n ._ Z _ _ .
moon: mnaiaafimacbm .303... .03 _ .uuoti .omw _ 03...: .03 _ 03...: bum _
E
masons? .mwo: baa _ .31. .3o _ .38.... b3 _ bum... .oqo _
58:5 .03 .omq _ b3 bug _ .emo bug _ bum .omq _
>ma _.:m_.:.i .oom _ _.:m_.::. bow _ _.:u_.:.:.. .25 _ #21:. box _
”8588 2:55 .31.: ._oo _ 63:... .So _
gamma—.8” ”Ed—v . c . . c
max gnaw—av room .5. _ #03 .EN _
Awomonasoo” 23.3 o o

 

 

5.8558 320 oE-£.§.o H 83.39 mom—do on @833 33 .1. N9 an .33 9.3%. 3. 5.8558 33° 03.353 .1. 52.3w. E. n “5“ an an?

93%— Nv. Era—Econ Baa oE-3=B.o .I. 2363. mm H mo. an Sea A380. 3. Ema—Econ Baa 03-3.53 n 33.3? QT an. nun“?— AZOQE 3.

208m” 0% u 25m :59 wmumSunEd 2.3: *uAbM 25¢: 1:..vaS. .25 «$238 88moQ mm amvoaaoaa £5. no @538— ESE—5..
.WESEm 2258: $5. 933%.: <meEa Ea game—48— :532.an Elma—om. :53 man so N03 8:9 9: 903 Ed 26 85 «BE—on Emu m<9

155

.520 Maw” 2.83588— rommmno ”06.0080: 32.0.0 0:82:65 CV wow—580:3 SE. 8 6:0» 0:0 0». 90 8922 0: >06 8: 56 08.00
8:008m 8: A8 Wow—880:8 «<5. :0 35:0:— ammwgza‘ 90:08:00 038on

 

 

 

 

 

32.0. a 32.0— N 32.0. m 32.0. :.
55:20 02 mm mm OW mm Um 0% mm E. 0% mm :m
m ._ z _ _ .
3:18. 9:80” 22 32.10: Qua: ._ G _ hem: ._ E _ _._ 3 ._~N — 6:3 .56 _
9088.80“ 3:303 . . o . . o . . c . . o
2:58. 6:830: :5: be: _ 68o boo _ Poo: boa _ .38 boo _
2:862. 3. 8:0 b3 .8: _ _.o:: .3: _ .3: .93 _ _.o:u baa _
2:862. om 3:883 bum .cae _ _.c:o .66: _ .o:m .3: _ row: .3: _
2:862 0:. 80: 06:90: .3: .3: _ .08 .3: _ b3 .3: _ .03 .3: _
mmnmmmoao: i 06:90: 60:: .23 _ .m:_ ._:o _ 6.3: .56 _ .m; ._:_ _
2:869. om mafia—06:90: roam bum _ row: bum _ 6:3 bum _ _.o:o .owu _
2:862. 0:. 808603 58m
8m0n60~ 8 "60 60:00:05 630...: .8: : 6:63.... .omm _ 63m... .8: _ Smut... .owm _
— I _. _ Z _ _ .
«<08 mam—80” 22 €2.55 63:: .3: _ _.Nm:_._ .Su _ rug... .Sm _ 6.36... .3: _
90:08:00“ £3585 . . c . . o . . o . . o
ads. 8820: cm sen—Aim 636:... bow _ 6:31.... bow _ 68: .08 _ 68w :8 _
n ._ Z _ _ .
moon: moaiaofimqoém .39.: .8: _ .38.: b3 _ .m _ m2... bum _ .313... bum _
E
9:088: .m:o_.i bum _ .mqw... .omm _ .391... .36 _ .36 bum _
80:80 _ .ccm .omm _ .30 .Su _ .03 b8 _ .3: .8: _
>m0 _ .861... box _ focus. .. bow _ rooms... .ocm _ 68:2... :3 _
”850:8 36:3 .3: .26 _ .31.. boo _
$03.88” ”:8: . . o . . c
w0x Q0883 Neat... .86 _ 93oz... .Em _
90:08:00” 255 . . o o

156

 

 

55:68: 8:: 0:63:80 H 53.39 :0m_.00 0:. 6.0038 3O n um“ :n :86 9:25 3. 5.8550: 8:: 03-358 u 32.3w. a». u wow :u :3:
AZRE NV. 5.8—Eco: 8:: 03-358 u 2:868. a». u mo. :1. :3: 9:80. 3. 5.8552: 8:: 03-353 u Snub—m. Eur. um. :n:ma: A330. 3.
2209 O” n 3% 8:9 mmumgama 0.3: 6A3“ :Ab: fivaS. .90 00:03:00 anamoQ :0 8088080 £5. 8. “.530:— aux—6:5.

_. ”888m 0303.6 £36 @8038. 81:20 8: 88810:. 8:38:08 81:65. 903 2.0 3 N03 00:0. 6:: 903 :8 25 00:0 088—0., :5: m<0.

Haw—o 9? 23:503.». rommmao womamwmoa Zena—m 008315 CV wow—532.8 £30 38.. 9o: :85. 3:2 :5: 353% 25 E
womvoaaoam £50 389 :5... =35. 363m.“ Aramaaoaoo magma—.5

 

 

 

 

 

zona— _ Zona— w 32.». u 33¢. a
<33? 0% mm a». 0% mm a». 0% mm E. OW mm mm
m ._ z _ _ .
3.58. MESH 22 Zion room .Ew _ rouq .Ew _ roam Lam _ Z 3 L3 _
932989 Ziav . . o . . o . . o . . o
3.58. 0:832. face .25 _ _ .25 .25 . _ boo .ooc _ _ .oco boo _
25573 cm 8:... f2.“ bum _ row; .30 — Po; bum _ .bww bu... _
2:862 om gnaw—.83 2:0 .3. _ .boc .oa _ Po; .8. _ ream b3 _
2.552. cm m8“. 9:96: fog baa _ _.cwo baa _ Z: w .oam _ raw. 9% _
wmammmnaoa i 0358: .bwm ._w_ _ row.“ ._ww _ rowo ._w_ _ _.cww ._ww _
2:869. a». «Bane—:53: .03 .8». _ .cmo .cwu _ boo .cww _ bow .oww _
25:62 2. 82:32.. =<§w
Smog—.2 5 :5 505255 .boo .oww _ roo— .oww _ roo— .owu _ roo— bwu _
_ I _. En _ _.
$2... mafia” 22 5.35% 63...... b3 _ 5&1... .omm _ .amwi. .omq _ 63...... .omm _
903638” «<03»ng . . o . . o . . o . . o
Hon: .3830: om Sea—”Em Foow .oow _ .bcw bow _ boo .oow _ .08 b8 _
EE
mafia moaiaofimnocvm : 3...... .owq _ Z 3...... .8.» _ ._._ 3...... .owq _ C 3...... .owq _
E
macaw—mo: #03 .03 _ fog... b3 _ rod .93 _ Pom.» .og _
58:5 rowq .o _ c _ #03 .o _ o w r85 .o w e _ ram .o _ e _
>mo .03.... boa _ .3“... b3 _ .39... bog _ .93.... b3 _
”830:8 9:8: room .03 _ .boo b3 _
Awomonaano” C155 . . o . . c
max 92:38 fwaw. .omc _ _.wum.. .26 _
€333.89 25—3 . . o o

 

 

Era—Econ 33o 05-3.88 u $92K“ nomaoa ow ”.339: 33 u wmw an .53 93am. 5. 5.8—Econ 3:0 03-3.53 u um? _w9 E. ".3“ 5".“on Ecua—
wv. 5.8558 320 02-353 u 394qu E. n won a u .33 93am— 3. 5.8558 330 03-358 u $989 an n ww“ a u 33 9.3%— 3.
Zeno...” O” u can... 339 mmufignma and... .25 836:8 gnomoJ. 5 5.62535 :50 38a 33a :85. 303%. 6A3“ ....A.o_“ 2...va3

3.58. 935. €01. mg“... ”3.338. 25 max 2.0 owSmoaoE anvaaaaa Ema—om. 0:0 om Ea .358 5 m2 8 s26 608:8 x mm Raga—ma”

157

qua—o b.wn guano—am. Fommmao ”amazon Zeno—m non—25% CV flaw—530:8 ($5 88a Saw. ram—E. $2.8 :5: 9.63% 25 E
”88333 £30 8.8 Ear. row—9 983% Ewan—6:8 OmSmon

 

 

 

 

32.n— _ 32?. w 32?. u 33a. 5

<m1w2o 0” mm a». 0% mm E. 0% mm 3. 0% mm 3.
m ._ Z _ _ .
338— mSEm” 22 28.18 .b: .25 _ .bE ._b_ _ f8» .25 _ .bqw ._bm _

anomanoaofl 52438 . . b . . b . . b . . o
2:58— 9:32. _ bbb bbb _ ~bbb bbb _ _ bbb bbb _ _bbb bbb .
25.52 o». 83 bow bum _ boa bmc _ bow bum _ bow bub _
2.553 c». 35:83 .buu b3 _ _bwb b3 _ .bug bo— b _bwm ba— _
2:563 ow m8“. 9:96: .03 b3 _ bob b3 _ .03 b3 _ bob b3 _
mmammwonoa E 2:58: .Sw: ._ 3 _ bum: ._ 3 _ bqw: ._ 3 _ bmu: ._ :w _
2:552 ow maggognaa .bwa bww _ ._bwm bww _ 7wa bww _ _bwo bww _
2552 om 33:2... 35m 8% .8... _ bx... .on _ .8“. .Su _ .8... .8w _

Homage. E 90 roan—53

$0? was? 22 £0135 wbnmi... Lou _ w.w_m:_.. ._bm _ wauI... ._bm _ wb. 1;... .25 _
Aroma—6:8” $9.523 . . b . . b . . b . . b
.38. 95:3: o». fiozasm _bbu bbw _ .bbw .bbw _ ~bbw bbw _ _bbw bow _

n ._ Z _ _ .
mofia moaiaaflmaocum bat... bum _ 9%.: bum _ mam}... bub _ 43...: bum _

E

macomnoa ba— bmb _ .3.» bu _ _ bub bub _ bqw bu _ _
98:5 bub... bw_ _ .omb bw. _ bum... bw_ _ .omb bw. _
>mo _ b _ b b3 _ _b _ b b3 _ _b_b b3 _ _b_b b3 _
”8538 95m: .cmm bmb _ bub bmb _
Awomonoaoo” C32; . . b . . b
max 9039.8 .bo. bb— _ roux bow _
93238” 33$ . . b b

 

 

5.8—Econ 33o nEéaE—R u $98.3 36.8 0». Wagon. 30 u wbw an 339381 3. 5.8558 82o 0E-£§ H mm... 59 E. Hub“ snubs Azan—
wv. Era—Econ 32o 03-353 u 393w“ E. u we“ a n .63 9.3%— 3. Ema—Econ 33o 03-353 n 3989 E. u w“ a u 33 A381 3.
228” O” u 8% 3:9 mmumgnwa 3.3. .35 83.38 gfimoQ mm nomvoaaga «$5 38a can. :85. Ban—mo. .vaM _.:..Abr :JAbb.

7:58. mum—Em. €01" mBEm. ”352.8. 25 max m3 Snamoaoa managing sins—am. Ono on :5 i=8 mm 8» 8 N26 608:8 x mm 895an

158

Hugo ff 3:538?— rommmmo ”mm—.033: Kean—m 00331.5 A: ”33333 «SE 95 2. 52.0 .838 o: mouaammmoa mow—a 25 A3
womuoaagfi 23. No.6 .803 o: “.3338: mom—a 903.38 95$on

 

 

 

 

32.». u gona— N Zona— u gona— A.
<mawzo 0% mm a». 0% mm 3. 0w mm E. Ow mm Um
m ._ z _ _ .

KNEE mSEm” 29 32.13 raw}... bow _ rmgfii b8 _ F393... .03 _ P338... b3 _
93.238” 22.123 . . o . . o . . o . . o
7.518. 9.330: Evo— _.:.. baa _ r2: .2 boo _ r2: 2. boo _ f8. 1. boo _
2552. o». 83 re“: .3.“ _ rouu b3 _ roam bun _ rcum .oua _
2:38“ on. $25383 roux bum _ #03 .03 _ fog .oma _ #39 b3 _
25562 cm $8 0353: .80 .omw _ but .omw _ bum .oau _ .ouo .omm _
mmnmmmnaoa i 0558: .mmm L : _ .mAu ._ .N _ .ma ._ I _ .mg ._ 5 _
2:89: o». manning—nan: Po: .8— _ roma .o~_ _ Po: .S— _ 7an bu— _
2.562 ow 333:. Sam .81: .8_ fi .0 5:... .SN _ .08.: .02 _ .21: .SN _
Swanson 5 Ga roam—.55
$9..» mafia” 2o" Sonar—w fumes... .omw _ #31... .23 — #30:... .omu _ 5qu .85 _
Awomoqgoo” €2,558 . . o . . c . . o . . o
H08. 95:5: om fan—SE .083... bow _ .081. .ocm _ .03... .oom _ boa .ccm _
n ._ z _ _ .
wear: unaiaofimaocvm .03 .03 _ _ .oom .omu _ .03 .93 _ foo» .onm _
E
manage: : $2. .03 _ fog b3 _ 55a... .. ban _ room. .95 _
58:5 .emo... .. .2 m _ .03: .o _ a _ .38.... .o _ m _ .93.... .o _ m _
>ma .03 bow _ boa bog _ b3 b3 _ .03 b3 _
”03938 Eng—av .booi... .oqw _ fwomfii bum _
903858” ”:35 o . . c
max gnaw—av .ma ~ ... .omm _ .mm _ _. bmu .
91.0838” 333 . o o

 

 

5.8—Econ 830 05-358 n amuhwfi nomao cm @0805 So H mm“ an no: 93%. C. Era—=53 :50 03-858 u $9.33 3. "we“ ~13: Eamo—

8. 555.com 330 03-353 n 3.53“ an n wow = u .33 A309“. 3. gnu—Econ Baa 03-358 n 30.3? E. H mm“ a n A“? 93%— 3.

Zofiom” or n 3% Snow mmumgama 2.33 JAbM :Ab : iiuAbS. .25 Ema—.38 ofiomoQ w Rave—50:8 £3. 83 moo...“ o: @3333: mow—o
._ 5535 203:5 £3. 98250:. <33?“ Ba omfiwoaoa 593032: $135? :58 Ed no mane 8:? mg :53 mm 25 on: mug—=2. Em: m5.

159

.000—0 0-? 700.8088. 000.80 ”00.00.08: 00000.0 00808.80 30 2080:0080 $.00 00 .0000 0:0 0.. 80.0 8.0080 000.00 0: 000.0008:
000.0 000 30 .w0000:00:$ £000 003 000.0 0: 008.0008: 000.0 900000000 0080on

 

 

 

 

.5000. 0 .5000. 0 3000. w .5000. 0.
$500.0 0” mm 0». OW mm 00. 0% mm 0». 0% mm Um
n ._ an _ _ .

2.01.0. 00080“ 20. 700.100 .0.0 .. .0 . .000 .30 . .000 ..8 . .000 .30 .
9000—0000” 7008.000 . . 0 . . 0 . . 0 . . 0
.5018. 08.000: . .000 .000 . . .000 .000 . .000 .000 . . .000 .000 .
2:809. 00. 00:0 .000 .000 . ...0. .000 . .000 .000 . ...00 .000 .
2:809. 00 0000.003 0.00.. .000 . .0002. .000 . .000i .000 . .300: .000 .
2:809. 0». 000.0 0.0.0.0: .000 .03 . .000 .000 . .000 .03 . .000 .03 .
000000000: <0 000.08: 300...: .50 . .5 .3... .50 . .0001... .50 . 500...: .50 .
2:802 00. 080008.08: .000 .08 . .000 .08 . .000 .08 . .000 .08 .
2:802 00. 80800.0 .880
80080.. 8 80 00:0000. 0 .03 .08 . .000 .000 . .03 .08 . .000 .000 .
«<00» 0880” 20" £000.00 0000...: .. .0 . 00.01. .. .0 . 0000...... .. .0 . ..0.0...... .. .0 .
90000080” $00080. . . 0 . . 0 . . 0 . . 0
0.0.0. 0:800: 0», 5.0880 .000 .000 . .000... .000 . .000. .000 . .000... .000 .
n ._ 2 _ _ .
M008. 098.00.403.80 .000»... .. .00. . .00 . 0...... .00. . .033... .00. . .0002... .03 .
E
m0:0000: .00... .000 . .05 .00. . .000... .000 . .000 .00. .
80080 .000 .08 . .000 .000 . .000 .000 . .000 .000 .
>00 .031... .000 . 0000...: .000 . .0002... .000 . . .0001... .000 .
”00.0800 8.00:. .000 .000 . .000 .000 .
900000000” 3:00 . . 0 . . 0
W00 9080.00 ..50 ..00 . ...3 .3 .
9000.030“ 700.00 . . 0 0

 

 

5.8.8000 800 08-00003 u 000.80. 000.00 00. 0.00008 0000 u 00. :u 0000 2000. .0. 5.8.8000 .000 08.30000 N 000.000. 00. "00. :u0000 90000.
00. 5.8.8000 .000 08-000000 u 000.000. 0». u 00. : n 0000 90000. 00. 008.8000 .000 08-000000 u 000.000. 0». n 00. : n 0000 90000. 00.

20000” OW n 0000 .000. mmu§00~0 0.40... 03.00. 3A0: ........0A.00.. .000 .000—0:00 00000000. .0 008000080 :00. N03 «090 0: 008.0008: 000.0.

. 58:80 0300000 <08 00.000008 «0000.0 0:0 00800000. 80008008 8080.00. 80.0 000 :0 N03 00:0. .w8 80.0 .0 0:0 00.. $0.00 000 080:9. 0.0: 080.

160

...00.0 0

70.0.8088. 000.000 ”000000.00 00000.0 00800000 ... ”000000080 000 0. .000. 0:0 0.000....0. .0 €0<0 . 00. N03 0.000....0. .0
£0<0 N. .0. ”000000080 000 N03 0.000....0~ .0 €0<0 .. 00.0. .000. 0:0 0.000....0. .0 £0<0 N. 5. ”000000080 00.000 0. .000. 0:0
0.000....0. 0: 00.0 €0<0 . 2.0 €0<0 N. .0. ”000000080 <0... 0. .000. 0:0 0.. .00 0:00.00 0: 0000 0:0 .20.. 000.00 8.00.00. 2.0 .0.
”000000008 03.80 N03 0.000....0. 0: £0<0 . 2.0 €0<0 0 ”000.0000 080....

 

 

 

>. .000. 000 1 N03 N03 1 >. .000. 0:0 00.00.0080 5.000....0. 20 >530... 70.00.00
5.0.00.0 0” mm 0.. 0” mm 0.. 0” mm 0.. 0” mm 0..
m .: Hm . _ — _ _.
2.0.00.0 ”0.00 .305 .50 . .00. ... .000. . .0001... 005 . .3031... 000 .
5000.80
702...... 9.8.8: . .00 . .00. . . .00 . .00. . ..00. .. .. .000 . ..000 .000 .
00.0. 00.000: 0.. $00.80 .000 .000 . ..000 .000 . ..000 000 . .003 000 .
000080.80
000.00. 0..... 0000.0: ..53 .000 . .000... .000 . .035 .005 . .000 ..00 .
....<.00 3.0000088.
20800.. 0.. 80800.0 ..<80 1....
800.00.. 8 .00 0000000. 0 ..000 .000 . .00.. 000 . ..00. 00. . ... .0 000 .
E
00000.8: .0. 0..... ..00. . .00 . . ..5 . .000 00.. . .00. ...... .000 .
80080 .000 .000 . .000 .000 . .0 . .... 000 . .035 0.0 .
>00 .0001. 0.0 . .0031... 0.0 . .0033... .0.. . .0002... 000 .
”00.00000 3000:. .300 0.0 . .20... ..00 . .300 ..00 . .000 03 .
”00000000. ”03.. . . 0 . . 0 . . 0 . . 0
00.. 9080.0. .30 .000 . .30 ..00 . .030 ..00 . 02003:... .000 .
90.000000. 000.0. 0 0 0 0

 

 

0.0000000 ..0..0 00.00020 N 030.000. 000.00 0.. ..000008 u 00. : n 0000 0000000008 .<0..0 00000..
20.00. 0” n 0000 .000. mmu080020 00.0... ....A00. ......A0.. 1......A00.. ...00 ..0..0..0:00 00.00000. .0 0000000080 00<.:0 N03 0.000....0. 8 00.0 $000 . 0:0
£0<0 N

161

.320 o

3:333». rommmao Womaommmos Kama—m Oovalbm 3v Wamwoaaoam 88n— Eow. row—9 963% 5 £96 _ Ea Sm<a N. E
”33338 88a 99a ram—E 605$ 958% E €98 H 25 296 N. E Womvosaga i5 88a 90: 53.9 wanna 5 230 M. $3
”83335 23o 88a :85. cone. 5 £95 7 25 A3 Wamvosmga 2:80 252$ o». mew—68a raw—9 2mm 33ng Ba A3
womuoaaoam 88a Emma row—E 639. E»: 368% E «<98 _ Ea €m<o N Qwamanonoo 9.2.3

 

 

 

 

 

 

mafia 35:53 was.“ 90.3.. ><o3ma~ wane. ... «<36 N 9032 ... €36 _ Emma-i
<33? Ow mm a». on mm an 0% mm a». or mm a». Ox mm an
u :. nm . _ — _ _.
3:52... ”38 magi... boo _ .muqti .oqw _ .381... .03 _ bun... bog _ 68...... .03 _
556:2.
751$. 6:320: room: .25 _ .682: boo _ ..oo... .25 _ ..oo. .08 _ roe—2:. boo _
05.38: o». 835% box bi _ bow... boa _ .03 boa _ bow bi _ boo: cog _
manna-8.2.
5i om 8:89 £9 2% .5 _ a: ..a _ .23 ..3 _ .8. .5. _ SM .5 _
9:96:
Siam 24.39252:
2:33 a 33:3. .03 .8“ _ .ed .85 _ 88 SN _ b3 .93 _ .08 SN _
5 9o rocmaroa
E
Mace—no: hem: .oom _ boat. .03 _ dam... .omu _ hon... .oom _ .35.... boo _
58:5 .08 .owm ~ bone: .8. _ .cd .owq _ 2x: .93 _ .03 gm _
>mn re; .o_o _ rob—2:. .o_o _ #8.... .So _ r933. boo _ foo—2.... boo _
”33383153 ..08 ._ 3 _ .3. ._Nu _ .ouu ._ I _ Fog ._ G _ #qu ._ 3 _
92.238” WE»: o o o o . c
max gnaw—3 .08 .50 _ ..3. .50 _ .Eu .5... _ r03 ..NN _ fume ..wa _
930838” :38 o o o o o

 

 

5.8—Econ Baa $1-355" u 3o. $9 9098 cm @0308 u 3“ a n 523 A<m=a 083v.

229..” Or n 3% Each wmumgama 2.8: ...ufloM ......Ab : ...:uAbS.

dz 833:8 gnome—.2 5 8302.38 «83 9o: :85. 3:2 Em: 3.5—mo E can. €m<o _ 25 :33 N.

162

.365 8

ZEEEBE bongo Wamaommmoa Zena—m 003335 CV Womuobaaaa 23. can 2. 808 898 o: @8833: moan 3 230 r
9: N08 808 E 298 N“ E ”83338 SE. N03 808 on nobnommmos moan 5 23a 7 9: can 3 :88 838 5 €w<o N. A3
Wamvosaga 23. can on 808 82.8 o: mauanmmmon mom—o 5 g9 «.55 _ Ba €m<m N. A3 Wamvoaaoam 2E. 25 9. :53 moo—.8
o: @8835: mom—a 335m. 85 A3 ”$635058 2:: N26 258 o: @3885: woman 5 won: 238 Qwamaqgoa «3:3.

 

 

 

 

 

>. .3: 25 1 No.6 No.6 I. >9 .893 can muaoanSo 52.333: 2: >310: 2:qu
SEN—Em Ow mm a». or mm 3. 0w mm mm Ow mm D».
n 5 nm . _ — _ _.
3:53." No.8 .08. .3. _ .9: .oqu _ has. bow _ .35.: .ofl _
9.252.
2318. 0:832. _.ooo .ooo _ rooo .ooo _ :5— .ooo _ ..ooo .ooo _
Hon: 95:8: om soawmsm boo bi _ ..ooo .oon _ be.» .26 _ bow: bow _
manna—3:2.
r96. ow 8:82 £3. 2:58: rowo .Eo _ .mi .30 _ .03 .Ga _ .39.... L3 _
53..“ >aqn=moEoE
2532 a. 3033; .25m .82 .8» _ .08: .8“ _ .ma... .8. _ .96. SA _
Swanson S 90 705905
E
macgaoa :3 .oqw _ ..omm bun _ _. _ .m boa _ .83 .03 _
38:5 boo .owc _ ..Su .80 _ .03 ... .owa _ bu _ .. .8. _
>mn ..Sa .So _ ..89... .So _ ran—... bow _ rods... b3 _
”832.8 enagv :3 ..NN _ _.o_ m ..NN _ 5.3.... .Em _ :mc .23 _
gonna—.8” 555 . . o . . o . . o . . c
max 92.5.3 .3»: .EN _ .mmm ..wn _ bunt: ._ 3 _ :3 b3 _
Awnmoasoo” 73.8 . . c . . o . . o . . o

 

 

5.8588 330 21-3ch H 80.39 cannon cm @5303 u no“ a u an; Asa—E 088V.
208m” 0% n 3% 8:9 mmnagafia and: *VAbm“ ......Ab: .....iuAbS. .55 Rmoaasoo 88mg! a 8.62538 3.35 83 moo...“ o: “.3338: mom—o 9
con. €30 _ 25 £96 N. womaaaoo Ea max Ed 88m018_ Emosonoaosmv 5558303 $53.3. 0.5 om 9a <wEom mm mm. 8 No.6 $850 w. mm 39533

163

 

W

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

amae (‘H’ i) to ask for reliance on the indulgence or
the goodwill of one another (Chapter 1)

bosei (fit-E) motherhood (Chapter 2)

bunmei kaika (Xflfififi‘lt) enlightenment and civilization (Chapter
2)

dekasegi (lilfi €) to migrate to other places of Japan to
supplement their income (Chapter 2)

emaki (fig) picture scrolls (Chapter 2)

Fukoku Kyohei (3.333%) one of the slogans in the Meiji period
indicating to enrich the country and
strengthen the military

eta (3;) pariahs of society, outcaste (Chapter 2)

goke (fl?) widow (Chapter 2)

ie (3) the traditional stem family (Chapter 1)

ie seido (gfillli) the system of primogeniture legally
recognized in the Meiji Civil Code in
1898 (Chapter 2)

ikigai (E é’b‘b‘) purpose of life (Chapter 7)

kazoku (2E) a term invented in the nineteenth
century to deal the Western idea of the
nuclear family in Japan (Chapter 2)

kodokushi (HMQR) solitary deaths (Chapter 7)

kogiroku (iifi) the popular moral writings describing
the importance of filial care of the
elderly (Chapter 2)

koseki (13$?) birth certificate, a household registrar
(Chapter 2)

kotobuki taisha (fiiflfl) women leaving jobs due to marriage

(Chapter 2)

 

mabiki (MEI 3‘)

abortion and infanticide, especially
when too many children were conceived
or born (Chapter 2)

 

Meiji iroha karuta (3533 L‘ 6 lifJ‘ 5 1’:)

a sort of traditional and popular
Japanese card game utilizing proverbs
(Chapter 1)

 

mokkan (*m)

wooden tablets (Chapter 2)

 

 

Oite futatabi chigo ni naru
(ttvczli'filztxé).

 

popular saying means that people
become a child again as they become
the old (Chapter 2)

 

164

 

 

 

 

 

 

 

 

Oitewa ko ni shitagae One of proverbs in Meiji iroha karuta

(tint 11%:1191) teaching to obey your children when
you are old (Chapter 1)

on (E) limitless obligations (Chapter 2)

otoji (XII E) grand toji, managed productive
enterprises within their independent
residences (Chapter 2)

oyabun and kobun (fifi¥fi) the lord-vassal relationship (Chapter 2)

seisai (IE!) lawful wife (Chapter 2)

sengyo-shufu (##Ifi) full-time homemakers (Chapter 2)

ubasute-yama ( 9??? 'C LU)

 

Umeyo, fuyaseyo (EU) J: fi‘bfi’ ck)

the mountains described in old tales
where the young left the elderly due to
financial difficulties and burden of care
work (Chapter 2)

 

 

shasekishfi (5475*), one of the slogans in the Meiji period
indicating, to give birth and increase the
population

shufu (Elk) collection of Buddhist tales (Chapter 2)

 

shumon aratame-cho (fiFfiafl‘E)

housewife (Chapter 2)

 

setsuwa (mtg)

religious and household registration
established to suppress Christianity
(Chapter 2)

 

terakoya (#45)

old tales (Chapter 2)

 

popular folk schools in the Tokugawa
period (Chapter 2)

 

toji (n a)

matron, mistress, housekeeper played a
major role in rural society, managing
agricultural enterprises and supervising
labor (Chapter 2)

 

wa (*U)

in the circle of harmonious integration

 

 

zekke (fig)

 

extinction of a family lineage

 

165

 

BIBLIOGRAPHY

Adelmann, P.K. 1994a. "Multiple Roles and Physical Health Among Older Adults:
Gender and Ethnic Comparisons." Research on Aging 16(2): 142-166.

—. 1994b. "Multiple Roles and Psychological Well-Being in a National Sample of
Older Adults." Journal of Gerontology 49:8277-8285.

Agresti, A. 1996. An introduction to categorical data analysis. New York: John
Wiley.

Antonucci, T.C.and H. Akiyama. 1995. "Convoys of Social Relations: Family and
Friendships within a Life Span Context." Pp. 355 - 372 in Handbook of Aging and
the Family, edited by R. Blieszner and V.H. Bedford. Westport, Conneticut:
Greenwood Press.

Asai, M.O.and V.A. Kameoka. 2005. "The Influence of Sekentei on Family
Caregiving and Underutilization of Social Services among Japanese Caregivers."
Social Work 50(2):]11-118.

Atoh, M., V. Kandiah, and S. Ivanov. 2004. "The Second Demographic Transition in
Asia? Comparative Analysis of the Low Fertility Situation in East and South-East
Asian Countries." The Japanese Journal of Population 2(1):42- 75.

Barnett, R.and J. Hyde. 2001. "Women, Men, Work, and Family: an Expansionist
Theory." American Psychologist 56(10):781-196.

Barnett, R.C.and N.L. Marshall. 2001. "Women, Men, Work, and Family: An
Expansionist Theory." American Psychologist 56:781-796.

Barnett, R.C., N.L. Marshall, and J .H. Pleck. 1992. "Men's Multiple Roles and Their
Relationship to Men's Psychological Distress." Journal of Marriage and the Family
54(2):358-367.

166

Beckett, M., N. Goldman, M. Weinstein, I.-F. Lin, and Y.-L. Chuang. 2002. "Social
Environment, Life Change, and Health Among the Elderly in Taiwan. " Social
Science and Medicine 55:191-209.

Benedict, R. 1946. The Chrysanthemum and the Sword: Patterns of Japanese
Culture. Cambridge, MA: The Riverside Press.

Berkman, L.F., T. Glass, 1. Brissette, and TE. Seeman. 2000. "From Social
Integration to Health: Durkheim in the New Millenium." Social Science and
Medicine 51 1843-857.

Brissette, I., S. Cohen, and TE. Seeman. 2000. "Measuring Social Integration and
Social Networks." Pp. 53-85 in Social Suppot Measurement and Intervention: A
Guide for Health and Social Scientists, edited by S. Cohen, L.G. Underwood, and
EH. Gottlieb. New York: Oxford University Press.

Brown, J .W., and J. Liang, N. Krause, H. Akiyama, H. Sugisawa, T. Fukaya. 2002.
"Transitions in living arrangements among elders in Japan: does health make a

difference?" Journal of Gerontology B: Psychological Sciences and Social Sciences.
57 (4): 8209-20.

Callahan, C.M.and F.D. Wolinsky. 1994. "The Effect of Gender and Race on the
Measurement Properties of the CES-D in Older Adults." Medical Care 32(4):341-
356.

Case, A.and C. Paxon. 2005. "Sex Differences in Morbidity and Mortality."
Demography 42(2): 189-214.

Chandola, T., P. Martikainen, M. Bartley, E. Lahelma, M. Marmot, S. Michikazu, A.
Naserrnoaddeli, and S. Kagamimori. 2004. "Does Conflict Between Home and Work
Explain the Effect of Multiple Roles on Mental Health? A Comparative Study of
Finland, Japan, and the UK." International Journal of Epidemiology 33(884-893).

Choi, N.G., J .A. Burr, M. Caro, and F .G. Caro. 2007. "Formal and Informal
Volunteer Activity and Spousal Caregiving Among Older Adults." Research on
Aging 29:99-124.

167

Chrouser, C.J.and A.-C.D. Ryff. 2006. "Multiple Roles and Well-Being:
Sociodemographic and Psychological Moderators." Sex Roles 55:801-815.

Cohen, ES. 2005. "Disability." Pp. 102-111 in Encyclopedia of Ageism, edited by
EB. Palmore, L. Branch, and D.K. Harris. New York: The Haworth Pastoral Press.

Cohler, B.J.and K. Altergott. 1995. "The Family of the Second Half of Life:
Connecting Theories and Findings." Pp. 59 - 94 in Handbook of Aging and the
Family, edited by R. Blieszner and V.H. Bedford. Westport: Greenwood Press.
Coontz, S. 2000. The Way We Never Were: American Families and the Nostalgia
Trap. New York: Basic Books.

Cornell, LL. 1987. "Hajnal and the Household in Asia: A Comparativist History of
the Family in Preindustrial Japan, 1600-1870." Journal of Family History 12(1): 143-
162.

—. 1996. "Infanticide in Early Modern Japan? Demography, Culture, and Population
Growth." The Journal of Asian Studies 55(1):22 - 50.

Comman, J .C., N. Goldman, D.A. Glei, M. Weinstein, and M.-C. Chang. 2003.
"Social Ties and Perceived Support: Two Dimensions of Social Relationships and
Health among the Elderly in Taiwan." Journal of Aging and Health

15:616-644.

Crimmins, E., S. Reynolds, and Y. Saito. 1999. "Trends in Health and Ability to
Work among the Older Working-Age Population." Journal of Gerontology Series B:
Psychological Sciences and Social Sciences 54(1):S31-S40.

Crimmins, E., Y. Saito, and S. Reynolds. 1997. "Further Evidence on Recent Trends
in the Prevalence and Incidence of Disability among Older Americans from Two
Sources: the LSOA and the NHIS." Journals of Gerontology Series B: Psychological
Sciences and Social Sciences 52(2):SS9-S71.

Davis, K.and J. Blake. 1956. "Social Structure and F ertiltiy: An Analytic
Framework." Economic Development and Cultural Change 4(3):211-235.

Durkheim, E. 1951. Suicide: A Study in Sociology. New York: The Free Press.

168

Dixon, R. B. 1978. "Late Marriage and Non-Marriage as Demographic Response:
Are They Similar?" Population Studies, 32 (3): 449-466.

Easterlin, R.A. 1995. "Economic and Social Implications of Demographic Patterns."
Pp. 73 - 89 in Handbook of Aging and the Social Sciences, edited by RH. Binstock
and L.K. George. San Diego: Academic Press.

Elder, G.H.J. 1985. Life Course Dynamics: Trajectories and transition: 1968 - I980.
Ithaca: Cornell University Press.

Fackler, M. 2007. "In Japan, Rural Economies Wane as Cities Thrive." in The New
York Times. New York.

Farris, W.W. 1985. Population, Disease, and Land in Early Japan, 645-900.
cambridge, MA: The Council on East Asian Studies, Harvard University, and the
Harvard-Yenching Institute, and disributed by the Harard University Pres.

Fredman, L., J. Magaziner, R. Hebel, W. Hawkes, and SI. Zimmerman. 1999.
"Depressive Symptoms and 6-Year Mortality among Elderly Community-Dwelling
Women." Epidemiology 10(1):54-59.

Fujiwara, Y., S. Shinkai, S. Kumagai, H. Amano, Y. Yoshida, H. Yoshida, H. Kim,
T. Suzuki, T. Ishizaki, S. Wataname, H. Haga, and H. Shibata. 2003. "Changes in
TMIG-Index of Competence by subscale in Japanese urban and rural community

older populations: Six years prospective study." Geriatrics and Gerontology
International 3 :S63-S68.

F ukuda, Y., K. Nakamura, and T. Takano. 2005. "Municipal Health Expectancy in
Japan: Decreased Healthy Longevity of Older People in Socioeconomically
Disadvantaged Areas." BMC Public Health 5(5).

Fuse, T. 1975. "Japan's Economic Development: Success, Stress, and Prospects for
the Future." Pp. 24 - 40 in Modernization and Stress in Japan, edited by T. Fusé. The
Netherlands: E. J. Brill, Leiden.

Garson, D.G.2006. "Logistic Regression." Available online at
http://www2.chissncsu.edu/garson/PA765/lgistic.htm. Accessed on December 19,
2007.

169

Garstka, H.M., T. Shaner, and S. Strahm. 1994. "Stereotypes of the elderly held by
young, middle-aged, and elderly adults." Journal of Gerontology: Psychological
Siences 49.

Glasgow, N. 2000. "Transportation Transitions and Social Integration of
Nonmetropolitan Older Persons." Pp. 108-131 in Social Integration in the Second
Half of Life, edited by K. Pillemer, P. Moen, E. Wethington, and N. Glasgow.
Baltimore: The John Hopkins University Press.

—. 2004. "Healthy Aging in Rural America." Pp. 271-281 in Critical Issues in Rural
Health, edited by N. Glasgow, L.W. Morton, and NE. Johnson. Ames, IA:
Blackwell Publishing.

Goldman, N. 1993. "The Perils of Single Life in Contemporary Japan." Journal of
Marriage and the Family 1993(55): 191-204.

Goldman, N., S. Takahashi, and Y. Hu. 1995. "Mortality Among Japanese Singles: A
Re-Investigation." Population Studies 49:227-239.

Goode, W.J. 1960. "A Theor of Role Strain." American Sociological Review
25(4):483-496.

Goto, M. 2002. Chfisei Kuge no Ie to Josei (1e and Women in Chfisei - AroundIZth
to 16th Century). Tokyo: Yoshikawa Hirofumi Kan.

Gove, W.R. 1973. "Sex, Marital Status, and Mortality." The American Journal of
Sociology 79(1):45-67.

Harnabata, M.M. 1990. Crested kimono. Ithaca: Cornell University Press.

Hanley, SB. 1972. "Toward an Analysis of Demographic and Economic Change in
Tokugawa Japan: A Villege Study." The Journal of Asian Studies 31(3):515 - 537.

—. 1985. "Family and Fertility in Four Tokugawa Villages." Pp. 196 - 228 in Family
and Population in East Asian History, edited by SE. Hanley and AP. Wolf.
Stanford: Stanford University Press.

170

Hareven, T.K. 1994. "Aging and Generational Relations: A Historical and Life
Course Perspective." Annual Review of Sociology 20:437-461.

—. 1995. "Chapter 2: Historical Perspective on the Family and the Aging." Pp. l3 -
31 in Handbook of Aging and the Family, edited by R. Blieszner and V.H. Bedford.
Westport: Greenwood Press.

Harris, P.B.and S.O. Long. 1999. "Husbands and Sons in the United States and
Japan: Cultural Expectations and Caregiving Experiences." Journal of Aging Studies
13(3):241-267.

Harrison, E.G. 1999. "Strands of Complexity: The Emergence of Mizuko Kuvo in
Postwar Japan." Journal of the American Academy of Religion 67(4):769-796.

Hashimoto, A. 1996. The Gift of Generations: Japanese and American Perspectives
on Aging and the Social Contract. New York: Cambridge University Press.

—. 2000. "Cultural meaning of "security" in aging policies." Pp. 19-27 in Caring for
the Elderly in Japan and the US. : Practices and policies, edited by SC. Long.

Hashizume, Y. 2000. "Gender Issues and Japanese Family-Centered Caregiving for
Frail Elderly Parents or Parents-in-Law in Modern Japan: From the Sociocultural and
Historical Perspectives." Public Health Nursing 17(1):25-31.

Hayami, A. 2001. Rekishi Jinko-gaku de Mita Nihon (Analysis of of Japan through
Historical Demography). Tokyo: Bubgei Syunju.

Hayami, A.and S. Kurosu. 2001. "Regional Diversity in Demographic and Family
Patterns in Preindustrial Japan." Journal of Japanese Studies 2(2):295 - 321.

Hazan, H. 1992. Managing change in Old Age: The Control of Meaning in an
Institutional Setting. Albany: State University of New York Press.

Holcombe, C. 1997. "Ritsuryo Confucianism." Harvard Journal of Asiatic Studies
57(2):543-573.

171

Horwitz, A., H. White, and S. Howell-White. 1996. "Becoming Married and Mental
Health: A Longitudinal Study of a Cohort of Young Adults." Journal of Marriage
and the Family 58(895-907).

Hu, Y.and N. Goldman. 1990. "Mortality Differentials by Marital Status. An
International Comparison." Demography 27:233-250.

Hummer, R.A., J. Pacewicz, S.-c. Wang, and C. Collins. 2004. "Health Insurance
Coverage in Nonmetropolitan America." Pp. 197-209 in Critical Issues in Rural
Health, edited by N. Glasgow, L.W. Morton, and NE. Johnson. Ames, IA:
Blackwell Publishing.

Idler, E.and Y. Benyamini. 1997. "Self-Rated Health and Mortality: A Review of
Twenty-Seven Community Studies." Journal of Health and Social Behavior
38(1):21-37.

Ikeda, A., H. 180, H. Toyoshima, Y. Fujino, T. Mizoue, T. Yoshimura, Y. Inaba, A.
Tamakoshi, and J .8. Group. 2007. "Marital status and mortality among Japanese men
and women the Japan Collaborative Cohort Study." BMC Public Health 7(73).

Ishizaki, T. 2004. "Active Life Expectancy and Predictors for Maintaining

Functional Independence among Older J apanese." Geriatrics and Gerontology
International 4:8132-8134.

Ishizaki, T., I. Kai, and Y. Imanaka. 2006. "Self-Rated Health and Social Role as
Predictors for 6-year Total Mortality among a Non-Disabled Older Japanese
Population." Archives of Gerontology and Geriatrics 42(1):91-99.

Jannetta, A.B.and S.H. Preston. 1998. "Two Centuries of Mortality Change in
Central Japan: The Evidence from a Temple Death Register." Pp. 79 - 98 in
Japanese Economic History 1600 - 1960, edited by M. Smitka. New York: Garland
Publishing, Inc.

Johnson, E. 1964. "The Stem Family and Its Extension in Present Day Japan."
American Anthropologist:839-851 .

172

Johnson, NE. 2004. "Spatial Patterning of Disabilities among Adults." Pp. 27-36 in
Critical Issues in Rural Health, edited by N. Glasgow, L.W. Morton, and NE.
Johnson. Ames, IA: Blackwell Publishing.

Kandel, D.B., M. Davies, and V.H. Raveis. 1985. "The Stressfulness of Daily Social
Roles for Women: Marital, Occupational and Household Roles." Journal of Health
and Social Behavior 26:64-78.

Kaneda, T.and J .M. Raymo. 2003. "Trends in the Quantity nd Quality of Life at
Older Ages in Japan." Pp. 147-175 in Demographic Change and the Family in
Japan's Aging Society, edited by J .W. Traphagan and J. Knight. New York: State
University of New York Press.

Kano, M. 2000. "Bosei-shugi to Nationalism (Motherhood and Nationalism)" Pp.
189-215 in Contemporary Sociology 19: Family Sociology, edited by S. Inoue, C.
Ueno, M. Oosawa, M. Mita, and S. Yoshimi. Tokyo: Iwanami-shoten.

Kasugai, N. 2004. Kaigo to Lifestyle no Shakaigaku (Sociology of Care and
Lifestyle). Kyoto: Sekaishiso-sha.

Katagiri, K. 2007. "Teinen Taishokusha no Syakaisanka ni Kakawaru 3tsu no
Shikousei (Analyses of retired people regarding reasons for participating in socal
activities and groups)" Presented at The Japanese Society of Social Phychology,
Tokyo, Japan.

Kawano, S. 2003. "Finding Common Ground: Family, Gender, and Burial in
Contemporary Japan." Pp. 125-144 in Demographic Change and the Family in
Japan 's Aging Society, edited by J .W. Traphagan and J. Knight. Albany: State
University of New York Press.

—. 2005. Ritual Practice in Modern Japan: Ordering Place, People, and Action.
Honolulu: University of Hawai'i Press.

Kikuzawa, S. 2000. "Multiple Roles and Health in Cross-Cultural Perspective:
Elderly Men and Women in the United States and Japan." Sociology, Indiana
University.

173

—. 2006. "Multiple Roles and Mental Health in Cross-Cultural Perspective: The
Elderly in the United States and Japan." Journal of Health and Social Behavior
47(1):62-76.

Kito, H. 2000. "Jinsei 40-nen no SekaisEdojidai no Shussei to Shibou (Life
Expectancy at Birth at 40 Years: Births and Deaths in Edo Period)" in Nihon Jinrui-
Gakkai Symposium.

—. 2001. Jinkou kara Yomu Nihon no Rekishi (Demographic Analysis of Japanese
History). Tokyo: Kodan-sha.

Klinenberg, E2002. Heat Wave: A Sociological Autopsy of Disaster in Chicago.
Chicago: University of Chicago Press.

Knight, J. 2003. "Repopulating the Village?" Pp. 107-124 in Demographic Change
and the Family in Japan 's Aging Society, edited by J .W. Traphagan and J. Knight.
New York: State University of New York Press.

Knight, J .and J .W. Traphagan. 2003. "The Study of the Family in Japan: Integrating
Anthropological and Demographic Approaches." Pp. 3-23 in Demographic Change
and the Family in Japan's Aging Society, edited by J .W. Traphagan and J. Knight.
Albany: State University of New York Press.

Koyano, W. 1997. "Myths and Facts of Aging in Japan." Pp. 213-227 in Aging
Asian Concepts and Experiences Past and Present, edited by S. Forrnanek and S.
Linhart. Wien: Osterreichische Akademie der Wissenchafien.

Krause, N., B. Ingersoll-Dayton, J. Liang, and H. Sugisawa. 1999. "Religion, Social
Support, and Health among the Japanese elderly." Journal of Health and Social
Behavior 40(4):405-421.

Krause, N.and J. Liang. 1992. "Cross-Cultural Variations in Depressive Symptoms
in Later Life." International Psychogeriarics 4 (Supplement 2): 1 85-202.

Kumabara, R. 2000. "Kindai Kazoku to Kafucho-sei (Modern Family and the Ie
System)" Pp. 137-157 in Contemporary Sociology 19: Family Sociology, edited by
S. Inoue, C. Ueno, O. Masachi, M. Mita, and S. Yoshimi. Tokyo: Iwanami-shoten.

174

Kumagai, F. 1984. "The Life Cycle of the Japanese Family." Journal of Marriage
and the Family 46(1):]91-204.

—. 1995. "Families in Japan: Beliefs and Realities." Journal of Comparative Family
Studies 26(1):]35.

Kushner, H.and C.E. Sterk. 2005. "The Limits of Social Capital: Durkheim, Suicide,
and Social Cohesion." American Journal of Public Health 95(7):1 139-1143.

Kweon, S.-I. 1997. ""Supu no Samenai Kyori (A Distance That Does Not Allow the
Soup to Get Cold)": The Search for an Ideal in the Care of the Eldelry in
Contemporary Japan." Pp. 369-381 in Aging Asian Concepts and Experiences Past
and Present, edited by S. Forrnanek and S. Linhart. Wien: Osterreichischen
Akademie Der Wissenschafien.

Kyodo News. 2007. "Japanese women still long-lived." in The Japan Times. Tokyo.

Kyoto Prefectural Library and Archives. 2005. "Sogo Shiryoukan Dayori No. 145
(Kyoto Prefectural Library and Archives Report)" Kyoto: Kyoto Prefectural Library
and Archives.

Lee, Y.-J.and S. Hirata. 2001. "Women, Work, and Marriage in Three East Asian
Labor Markets." Pp. 96 - 124 in Women 's Working Lives in East Asia, edited by
MC. Brinton. Stanford: Stanford University Press.

Long, S.O.and P.B. Harris. 2000. "Gender and Elder Care: Social Change and the
Role of the Caregiver in Japan." Social Science Japan Journal 3(1):21-36.

Maeda, D. 2004. "Societal Filial Piety Has Made Traditional Individual Filial Piety
Much Less Importatn in Contemporary Japan." Geriatrics and Gerontology
International:S74-S76.

Maeda, D.and H. Ishikawa. 2000. "Ageing in Japan: Retirement, Daily Lives,
Pensions and Social Secmity." Pp. 113-132 in Ageing in the Asia-Pacific Region:
Issues, Policies and Future Trends, edited by DR. Phillips. New York: Routledge.

175

Manton, K., E. Stallard, and L. Corder. 1995. "Changes in Morbidity and Chronic
Disability in the US. Elderly Population: Evidence from the 1982, 1984, and 1989

National Long Term Care Surveys." Journal of Gerontology Series B: Psychological
Sciences and Social Sciences 50(4):Sl94-8204.

Martin, LG. 1989. "The Graying of Japan." Population Bulletin 44(2).

Matsui, T., T. Ohsawa, and M.-L. Onglatco. 1995. "Work-Family Conflict and the
Stress-Buffering Effects of Husband Support and Coping Behavior among Japanese
Married Working Women." Journal of Vocaional Behavior 47.

McLaughlin, D.K., C.S. Stokes, J.P. Smith, and A. Nonoyarna. 2007. "Differential
mortality across the United States: The influence of place-based inequality." in The
Sociology of Spatial Inequality, edited by L.M. Lobao, G. Hooks, and AR.
Tickamyer. Albany: State University of New York Press.

McLaughlin, D.K., S.C. Stokes, and A. Nonoyarna. 2001. "Residence and Income
Inequality: Effects on Mortality among US. Countries." Rural Sociology 66(4):579-
599.

Merton, R.K. 1968. Social Theory and Social Structure. New York: Free Press.

Mertens, W. 1994. "Health and Mortality Trends Among Elderly Populations:
Determinants and Implications." Liege, Belgium: International Union for the
Scientific Study of Population. Policy and Research Paper 3.

Ministry of Education Culture Sports and Science and Technology. 2005. Japan 's
Education at Glance 2005. Tokyo: Ministry of Health, Labour and Welfare.

Ministry of Foreign Affairs. 2007. "Social Security System: An Aging Society and
Its Impact on Social Security." in Japan Fact Sheet.

Ministry of Health Labour and Welfare.2003. "White Paper on the Labour Economy
2003." Available online at
http://wwwhakusyo.mhlw.go.ip/wpdocs/lggax200301/b0001.htrnl. Accessed on
January 20, 2008.

176

Moen, P. 2001. "The Gendered Life Course." Pp. 179-196 in Handbook of Aging and
the Social Sciences, edited by RH. Binstock and L.K. George. New York: Academic
Press.

Moen, P., Dempster-McClain, and R.M.J. Williams. 1989. "Social Integration and
Longevity: An Event History Analysis of Women's Roles and Resilence." American
Sociological Review 54(4):635-647.

—. 1992. "Successful Aging: A Life-Course Perspective on Women's Multiple Roles
and Health." The American Journal of Sociology 97(6): 1612-1638.

Moffett, S. 2005. "F ast-Aging Japan Keeps Its Elders on the Job Longer." The Wall
Street Journal.

Mori, K. 1993. Haka to Sousou no Shakaishi (Social History of Graves and Burial
Services). Tokyo: Kodansha.

Mori, M., Y. Nakashima, Y. Yamazaki, and H. Kurita. 2002. "Sex-Role Orientation,
Marital Status and Mental Health in Working Women." Archives of Women 's Mental
Health 5:161-176.

Nakane, C. 1967. Kinship and Economic Organization in Rural Japan. London: The
Athlone Press.

—. 1970. Japanese Society. Berkeley: The University of California Press.

Ochiai, E. 1996. The Japanese Family System in Transition: A Sociological Analysis
of Family Change in Postwar Japan. Tokyo: LTCB International Library
Foundation.

Ogawa, N. 2004. "Urban-Rural Differentials in Health Conditions and Labor Force
Participation among the Japanese Elderly." Geriatrics and Gerontology International
4(S60-S62).

Ogawa, N.and R.D. Retherford. 1993. "Care of the Elderly in Japan: Changing
Norms and Expectations." Journal of Marriage and the Family 55:585-597.

177

—. 1997. "Shifting Costs of Caring for the Elderly Back to Families in Japan: Will It
Work?" Population and Development Review 23(1):59-94.

Okada, K. 1998. "Redundancy Utilized: the Sconomics of Female Domestic Servants
in Pre-war Japan." Pp. 146 - 166 in Japanese Economic History 1600 - 1960, edited
by M. Smitka. New York: Garland Publishing, Inc.

Omran, AR. 1971. "The Epidemiologic Transition: A Theory of the Epidemiology
of Population Change." Milbank Memorial Fund Quarterly 49(4):509-538.

Oshio, T.and A.S. Oishi. 2004. "Social Security and Retirement in Japan: An
Evaluation Using Micro-Data." Pp. 399-460 in Social Security Programs and
Retirement around the World Micro-Estimation, edited by J. Gruber and DA.
Wise. Chicago: The Universit of Chicago Press.

Ota, K. 2006. "Syoshi Koreika Sekai-ichi ni (Japan Became Number 1 Aging
Society)" Pp. 3 in Asahi Shinbun. Tokyo.

Palmore, EB. 2005. "Stereotypes." Pp. 301-303 in Encyclopedia of Ageism, edited
by EB. Palmore, L. Branch, and D.K. Harris. New York: the Haworth Pastoral

Press.

Pillemer, K.and N. Glasgow. 2000. "Social Integration and Aging." Pp. 19 - 47 in
Social Integration in th Second Half of Life, edited by K. Pillemer, P. Moen, E.
Wethington, and N. Glasgow. Baltimore: The John Hopkins University Press.

Quandagno, J. 2005. Aging and the Life Course: An Introduction to Social
Gerontology. New York: McGraw Hill.

Raymo, J.M. 1998. "Later Marriages or Fewer? Changes in the Marital Behaviors of
Japanese Women." Journal of Marriage and the Family 60(4): 1023-1034.

Retherford, RD, N. Ogawa, and R. Matsukura. 2001. "Late Marriage and Less
Marriage in Japan." Population and Development Review 27(1):65-102.

178

Retherford, R.D., N. Ogawa, R. Matsukura, and H. Ihara. 2004. Trends in Fertility
By Education in Japan, 1966-2000: Nihon University of Population Research
Institute: Tokyo., East-West Center: Honolulu, and Statistical Bureau: Tokyo.

Retherford, R.D., N. Ogawa, and S. Sakamoto. 1996. "Values and Fertility Change
in Japan." N UPRI Reprint Series 73.

Robertson, J .F. 1995. "Grandparenting in an Era of Rapid Change." Pp. 243 - 260 in
Handbook of Aging and the Family, edited by R. Blieszner and V.H. Bedford.
Westport: Greenwood Press.

Rodgers, R.G., R.A. Hummer, and CB. Nam. 2000. Living and Dying in the USA:
Behavioral, Health, and Social Diflerentials of A dult Mortality. New York:
Academic Press.

Rogers, R.G., R.A. Hummer, and CB. Nam. 2000. Living and Dying in the USA:
Behavioral, Health, and Social DWrentials of Adult Mortality. San Diego:
Academic Press.

Ross, C., J. Mirowsky, and K. Goldsteen. 1990. "The Impact of the Family on
Health: The Decade in Review." Journal of Marriage and the Family 52:1059—1078.

Ross, C.E.and C.-L. Wu. 1995. "The Links between Education and Health."
American Sociological Review 60:719-745.

Rowe, J.and R. Kahn. 1997. "Successful Aging." Gerontologist 37:433-440.

Rozario, P.A., N. Morrow-Howell, and J .E. Hinterlong. 2004. "Role Enhancement or
Role Strain." Research on Aging 26(4).

Saito, O. 1992. "Infanticide, Fertility and 'Population Stagnation': The State of
Tokugawa Historical Demography." Journal of Asian Studies 4(2):369-381.

—. 2000a. "Introduction: The Economic and Social Aspects of the Family Life Cycle
in Traditional and Modern Japan." Continuity and Change 15(1):11-15.

179

—. 2000b. "Marriage, Family Labour and the Stem Family Household: Traditional
Japan in a Comparative Perspective." Continuity and Change 15(1): 17-45.

Schoeni, R.F., J. Liang, J. Bennett, H. Sugisawa, T. F ukaya, and E. Kobayashi. 2005.
"Trends in Old-Age Functioning and Disability in Japan: 1993-2002." Working
Paper Series: TRENDS Evaluating trends in old-age disability 5(3).

Schoppa, L.J. 2006. Race For the Exits: The unraveling of Japan '3 System of Social
Protection. Ithaca: Cornell University Press.

Senuma, Y. 2000. Cho’ju-shakai no ikigai sagashi (Finding ikigain in aging society).
Tokyo: Daimeido.

—. 2006. Kourei Yoka ga Chiiki wo Tsukuru (how the elderly spend the later life
course contributes to the community). Tokyo: Gakubunsha.

Shibata, H., H. Sugisawa, and S. Wataname. 2001. "Functional Capacity in Elderly

Japanese Living in the Community." Geriatrics and Gerontology International 1:8-
13.

Shinozaki, Y., E. Nishikawa, and T. Deguchi. 2005. Lessons fiom the Great Hanshin
Earthquke. Kobe: Hyogo Research Center for Quake Restoration.

Shiozaki, Y. 2005. " Shadows of creative reconstruction." Presented at Regional

Conference on Reconstruction and Development of Peasants' and Fisherfolks'
Livelihoods, Medan, Indonesia.

Shirahase, S. 2000. "Women's Increased Higher Education and the Declining
Fertility Rate in Japan." Review of Population and Social Policy 9:47-63.

Shryock, H.S.and J.S. Siegel. 1976. The Methods and Materials of Demography, Vol
1. Washington, DC: US. Bureau of the Census.

Smith, SR. 2000. "Land of the Rising Son? Domestic Organization, Ancestor
Worship, and Economic Change in Japan." Pp. 305-323 in Family, Religion, and
Social Change in Diverse Societies, edited by S.K. Houseknecht and J .G. Pankhurs.
New York: Oxford University Press.

180

Smith, J .P. and Kington, R. 1997. "Demographic and Economic Correlates of Health
in Old Age." Demography 34 (1): 159-170.

Sugano, N. 1999. Edo-jidai no Kokomono (F ilial Piety in the Eda-Period). Tokyo:
Yoshikawa Kobunkan.

Sugisawa, H., J. Liang, and X. Liu. 1994. "Social Networks, Social Support, and
Mortality Among Older People in Japan." Journal of Gerontology 49(1):S3-S l 3.

Taeuber, LB. 1958. The Population of Japan. Princeton: Princeton University Press.

Takagi, E., M. Silverstein, and E. Crimmins. 2007. "lntergenerational Coresidence of
Older Adults in Japan: Conditions for Cultural Plasticity." Journal of Gerontology
62(5):S330-S339.

Takeda, Y., I. Kawachi, Z. Yamagata, S. Hashimoto, S. Matsukura, S. Oguri, and A.
Okayama. 2006. "The Impact of Multiple Role Occupancy on Health-Related
Behaviors in Japan: Differences by Gender and Age." Public Health 120:966-975.

Takeshita, J ., K. Masaki, I. Ahmed, D.J. Foley, Y. Qing Li, r. Chen, D. Fujii, W.G.
Ross, H. Petrovitch, and L. White. 2002. "Are Depressive Symptoms a Risk Factor
for Mortality in Elderly Japanese American Men? The Honolulu-Asia Aging Study."
Am JPsychiatry 159:1127-1132.

Tamney, J .B. 1998. "Confircianism." in Encyclopedia of Religion and Society, edited
by W.H.J. Swatos. Walnut Creek, CA: AltaMira Press.

Tanaka, K.and N.E. Johnson. 2006. "What Japan Can Do To Push Its Longevity
Envelope." Population Reference Bureau.

Thang, LL. 2000. Generations in Touch: Linking the Old and Young in a Tokyo
Neighborhood. Ithaca: Cornell University Press.

Thoits, RA. 1991. "On Merging Identity Theory and Stress Research." Social
Psychology Quartely 54(101-112).

181

—. 1992. "Identity Structures and Pychological Well-Being: Gender and Marital
Status Comparisons." Social Psychology Quartely 55(236-56).

Thompson, CS. 2003. "Depopulation in Rural Japan: "Population Politics" in Towa-
cho." in Demographic Change and the Family in Japan '3 Aging Society, edited by
J .W. Traphagan and J. Knight. Albany: State University of New York Press.

Tipton, E.K. 2002. Modern Japan: A social and political history. New York:
Routledge.

Totman, CD. 1982. Japan Before Perry: A Short History. Berkely: University of
California Press.

Traphagan, J .W. 2003. "Contesting Coresidence: Women, In-Iaws, and Health Care
in Urban Japan." Pp. 203 -226 in Demographic Change and the Family in Japan 's
Aging Society, edited by J .W. Traphagan and J. Knight. Albany: State University of
New York Press.

—. 2004. The Practice of Concern: Ritual, Well-Being, and Aging in Rural Japan.
Durham, NC: Carolina Academic Press.

Troll, LE. 1995. "Foreword." Pp. xi - xx in Handbook of Aging and the Family,
edited by R. Blieszner and V.H. Bedford. Westport: Greenwood Press.

Ueno, C. 1998. "Shusseiritsu no Teika: Dare no Mondaika? (Declining Fertility:
Whose Problems Is It?)." Jinko mondai kenkyu 54(41-62).

Umberson, D., K. Williams, D.A. Powers, M.D. Chen, and A.M. Champbell. 2005.
"As Good as it Gets? A Life Course Perspective on Marital Quality." Social Forces
84(1):493-511.

van de Kaa, DJ. 1987. "Europe's Second Demographic Transition." Population
Bulletin 42(1):1-59.

Voydanoff, P.and B.W. Donnelly. 1999. "Multiple Roles and Psychological Distress:
The Intersection of the Paid Worker, Spouse, and Parent Roles with the Role of the
Adult Child." Journal of Marriage and the Family 61(3):725-738.

182

Walker, A.J., K.R. Allen, and LA. Connidis. 2005. "Theorizing and Studying Sibling
Ties in Adulthood." Pp. 167-190 in Sourcebook of Family Theory and Research,
edited by V.L. Bengtson, A.C. Acock, K.R. Allen, P. Dilworth-Anderson, and D.M.
Klein. Thousand Oaks: Sage.

Weeks, J .R. 2002. Population: An Introduction to Concepts and Issues. Belmont,
CA: Wadsworth/Thompson Learning.

Wethington, E., P. Moen, N. Glasgow, and K. Pillemer. 2000. "Multiple Roles,
Social Integration, and Health." Pp. 48 - 71 in Social Integration in the Second Half
of Life, edited by K. Pillemer, P. Moen, E. Wethington, and N. Glasgow. Baltimore:
The John Hopkins University Press.

Wickrama, K., R.D. Conger, F.O. Lorenz, and L. Matthews. 1995. "Role Identity,
Role Satisfaction, and Perceived Physical Health." Social Psychology Quartely
58(4):270-283.

Wu, Y. 2002. "Speaking of Life: Life Histories of the Institutionalized Elderly in
Kotobuki." Intersections: Gender, History & Culture in the Asian Context 7.

Wu, Z.H.and L. Rudkin. 2000. "Social Contact, Socioeconomic Status, and the
Health Status of Older Malaysians." The Gerontologist 40(2):228-235.

Yanagiya, K. 2001. "NIhon Kinse no Koreikaigo to Kazoku (Elder care and the
family in modern Japanese history)" Pp. 171-202 in Kaigo to Kazoku (Care and
Family), edited by E. Yamanaka, E. Takeyasu, H. Sone, and R. Shiraishi. Tokyo:
Waseda University.

Yasuda, M. 1992. Ohaka Ga Naito Shinema Senka (Can 't We Die Without Graves?)
Tokyo: Iwanami-shoten.

Yoshie, A. 2005. "Gender in Early Classical Japan: Marriage, Leadership, and
Political Status in Village and Palace." Monumenta Nipponica 60(4):437 - 479.

Zhang, Z.and M. Hayward. 2006. "Gender, the Marital Life Course, and
Cardiovascular Disease in Late Midlife." Journal of Marriage and the Family
68:639-657.

183

 

lljlfilljjllljljlfllj

    

111131511111

6