Filial expectations and social exchange patterns among older Taiwanese parents and their adult children

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Request accessible transcript

Transcript (if available)

Content FILIAL EXPECTATIONS AND SOCIAL EXCHANGE PATTERNS AMONG OLDER TAIWANESE PARENTS AND THEIR ADULT CHILDREN by Frances M. Yang A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (GERONTOLOGY) MAY 2004 Copyright 2004 Frances Margaret Yang R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. UMI Number: 3140578 Copyright 2004 by Yang, Frances M. All rights reserved. INFORMATION TO USERS The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. UMI UMI Microform 3140578 Copyright 2004 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 11 DEDICATION This is dedicated first and foremost to, Jesus Christ, the Greatest Teacher who ever lived, to whom I aspire to be more like each day in the academic pursuit of knowledge and scientific discoveries; To my grandparents, to whom I hope to be a crown for the aged; To my parents, To whom I will always honor and obey; To my brother, whom I have undeservingly been blessed with to be my closest friend through this walk of life; To the one, whom God has blessed me to love and grow old with; To the people of my beloved countries: Taiwan and America, to whom I am called to serve for the rest of my life . R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Ill ACKNOWLEDGEMENTS I remember when I was five years old I wanted to be both an author and detective. Twenty-six years later, I have finished writing a dissertation on my research investigating filial expectations and other motivations behind intergenerational social support patterns found within the Taiwanese family. As I reflect upon my own family, I have realized that even though I was bom and raised in the United States, the filial normative values found in Taiwan have been firmly ingrained in me. Among the four domains of motivations to provide support for older parents, which I have explored in this dissertation, I find that my family’s strongest motivator will be ideology. When my grandparents were still alive, I grew up in a multi-generational household both in the United States with my paternal grandparents and every other summer in Taiwan with my maternal grandparents. I found that my paternal grandparents relied heavily upon their children for support, while my maternal grandparents were the main supporters of their youngest adult children. Both sets of grandparents were bom in Taiwan, yet grew old and passed away in two different countries, but both had adult children who were willing to unconditionally express filial piety. Thus, setting the stage for my career in the field of gerontology and setting the foundation for this dissertation. But this joumey to this point of my life could not have been possible without the generosity and kindness of the following people. The words in the following pages of this R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. IV acknowledgement are only a fraction of the gratitude and appreciation that I have for each of them. My dissertation chair and mentor, Merril Silverstein, is the key reason that I have embarked on this career path in gerontology. I have been blessed with the privilege and honor of working with one of the finest scholars in gerontology for the past six years of my life, beginning my senior year as an undergraduate at USC. I want to thank him for believing in me and encouraging me to enter this Ph.D. program in gerontology, which has opened up unimaginable doors. I hope to be faithful in this pursuit of academic rigor and to make an impact in the aging field that will help eountless individuals for many generations to eome. The three other members of my dissertation committee, Vem Bengtson, Eileen Crimmins, and Timothy Biblarz, in order of acquaintance, have each played a significant role in my academic life. My first encounter with gerontology was through Vem’s “Family and Aging” class, which lead me to this life-changing decision to pursue a career in aging. It was also Vem’s “Theories of Aging” elass that taught me to write, think, and conduct research—using theory. This dissertation is a result of his painstaking editing and strong criticism of all my papers, qualifying exam answers, and dissertation drafts. I am eternally grateful for your guidance and for your vision for the University of Southern Califomia Longitudinal Study of Generations (LSOG) research project, with which I have had the opportunity to use the data under the direction of Merril Silverstein. Eileen has been my greatest inspiration, with her success as an academic, a wife, and a mother. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. V I hope to one day be like her and learn to balance life’s responsibilities like the way she does, while establishing many connections in research and scholarly exchange here and abroad. She has indeed been a key role model and an important contact for my opportunity to work at the Taiwan Provincial Institute for Family Planning (now called, Bureau of Health Promotion and Protection (BMP)) under the direction of Ming-Cheng Chang, Harvey Lin, and Yi-Li Chuang. I truly enjoyed that summer working in Taichung, Taiwan and hope to one day work there again with all my research associates whom I miss dearly. No words can truly express my gratitude for Eileen’s encouraging words, hugs, and advice during the times I felt so lost in the field of aging. But she has been the anchor for many of the students in conquering and withstanding the worst storms in the Ph.D. program. Tim taught me the humor behind statistics when I took his “Advanced Statistics” course. His optimism and teaching methods is a model that I follow in teaching research methods and statistics. I truly appreciate Tim’s willingness to sit on my dissertation committee and offer statistical advice and for listening to my finstrations. I want to, once again, thank each of my committee members for their inspiration, encouragement, and wisdom in helping me mold and refine this dissertation to what it is today. I want to thank both the BHP (Ming-Cheng Chang, Director) and the Population Studies Center at the University of Michigan, Ann Arbor (Albert I. Hermalin, P.I.) for providing all the waves of the Survey of Health and Living Status of the Middle Aged and Elderly in Taiwan. I want to acknowledge both institutions that have generously allowed me to use the data and the phenomenal research staff in R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. VI Taiwan that promptly answered my questions concerning the data. I would like to acknowledge funding of the survey project that was provided by a number of Taiwanese government agencies, namely the Department of Health, the National Health Research Institute (Grant No. DDO1-86IX-GR01S), and the Executive Yuan. The U.S. National Institute of Aging (Grant No. R37 AG07637) also provided supplemental funding. My grandfather, Lin Shien-Hua, a man who lived until he was 99 years old and a major influence on my life in pursuing a career in gerontology. I will never forget his ability to play table tennis, archery, and travel at such a ripe old age. I wish so much he could have lived to see me attain my Ph.D. degree, but I am grateful for his love and support throughout his lifetime. It is because of my grandfather that I fell in love with Ilha Formosa, I hope to make contributions like the way he did in the lives of the people he touched. His legacy will be passed on to many generations for the sacrifices he’s made on behalf of his family and the generosity he extended to those in need. My supportive and loving parents, Lloyd and Jing-Howe Yang, taught me to pursue God’s calling for me. They have always prepared me for life’s greatest challenges and given me the freedom to reach for the stars. I want to thank my mother for her unceasing prayers and words of wisdom for me throughout my life thus far. For listening to my sufferings, and reminding me to “rejoice in our sufferings, because we know that suffering produces perseverance; perseverance, character; and character, hope. And hope does not disappoint us, because God has R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. V ll poured out his love into our hearts by the Holy Spirit, whom he has given us (Romans 5:3-5).” I have been eternally blessed to have them as my parents, and for believing that I will one day make a great contribution to the people in this world. I want to thank my dearest brother, Ted, who is a constant reminder of living a life of Christian integrity. He is the only one on earth that truly understands and empathizes with me. Ted has offered me more sound advice and wisdom as a younger brother than any older sister could give. May God bless you richly in all your endeavors. I want to thank all my relatives, friends, co-workers, classmates, and my Christian brothers and sisters for their prayers throughout this long joumey. If it weren’t for each of them, I would not be able to conquer all the disappointments, heartaches, and trials that I have gone through. I look forward to many more years of laughter through the victories, triumphs, and joys that we will be experiencing together. I would like to share with my past, current, and future gerontology students that you have made a wise investment to enter this exciting field of the future. To all those I have yet to meet and who will ask me: “What is gerontology?” My answer will always be: “It’s the study of aging, not the study of old people.” Above all, I want to thank my Lord and Savior, Jesus Christ, for being the light on this path of life that often gets dark and lonely. May I use the gifts you have blessed me with to glorify you always. I am etemally indebted to His unconditional love, mercy, and grace for me. " I do not consider myself yet to take hold of it. But one thing I do: Forgetting what is behind and straining toward what is R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. V lll ahead, I press on toward the goal to win the prize for which God has called me heavenward in Christ Jesus" (Philippians 3:13-14). March 2004 Andrus Gerontology Center University of Southern Califomia R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. TABLE OF CONTENTS Page No. DEDICATION ii ACKNOWLEDGEMENTS, iii LIST OF TABLES ix LIST OF FIGURES xi ABSTRACT xii Chapter I. INTRODUCTION Purpose 1 Significance of the Problem 2 II. BACKGROUND OF TAIWAN RELATIVE TO OTHER ASIAN COUNTRIES Can Asian Families Continue to Support the Older Generation Despite the Consequences of Population Aging? 4 Demographic, Economic, and Social Changes in Taiwanese Families in Relation to Other Asian Countries 7 Uniqueness of Taiwan’s Geographical, Historical, and Cultural Impact upon Families 19 III. THEORETICAL BACKGROUND AND LITERATURE REVIEW What are Intergenerational Transfers? And Why are They Important? 28 Intergenerational Transfers in Taiwan 30 The Influence of Modernization on Intergenerational Transfers in Taiwan 31 The Process and Paths of Transformation Toward Modernization in Taiwan 42 Micro-Social Level Theories of Social Support Exchanges in Taiwan 46 Intergenerational Transfer Models 48 Statement of the Problem 5 5 Conceptual Framework 58 Research Questions 67 Hypotheses R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. IV. RESEARCH DESIGN AND METHODS Data Source 73 Measures 75 Overview of Statistical Analyses 102 V. RESULTS Forming the Latent Class Clusters for Directions of Social Support Transfers 115 Multinomial Logistic Regression 124 Emotional and Instrumental upward Transfer Class versus No Upward Transfer Class 126 Economic Upward Transfer versus No Upward Transfer Class 129 Full Upward Transfers versus No Upward Transfer Class 132 VI. DISCUSSION AND CONCLUSION Summary of Study Results 136 Limitations and Future Directions of the Study 137 Implications for Family Theory 152 Imphcations for Social Policy in Taiwan 154 REFERENCES 157 APPENDICES Appendix A: Odds Ratios for Multinomial Logistic Regression for Types of Support Transfers (N=9477) 163 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. IX Table 1. LIST OF TABLES Page No. The Intergenerational Social Support Models for Older Taiwanese Parents Based upon Economic, Instmmental, and Emotional Support 60 Table 2. Description of Models of Motivation for Intergenerational Transfer within the Matrix of Domains of Motivations by Direction of Transfers 66 Table 3. Table of Independent Variables Used in the Analysis 78 Table 4. Table of Dependent Variables Used in the Analysis 83 Table 5. Frequency Distribution and Percentages of Support Indicators 84 Table 6. Distribution and Means of Respondent Sociodemographic Characteristics in 1999 86 Table 7. Likelihood Ratio Chi-Squares for Selected Models 118 Table 8. Goodness-of-fit Statistics for the Four-Class Model 119 Table 9. Conditional Latent Class Probabilities for a Four-Class Model 120 Table 10. Simplified Table of Odds Ratios from Multinomial Logistic Regression of Being in Emotional and Instrumental Upward Transfer vs. No Upward Transfer for Models of Motivation (comparison group) 129 Table 11. Simplified Table of Odds Ratios from Multinomial Logistic Regression of Being in Economic Upward Transfer vs. No Upward Transfer for Models of Motivation (comparison group) 130 Table 12. Simplified Table of Odds Ratios from Multinomial Logistic Regression of Being in Full Upward Transfer vs. No Upward Transfer for Models of Motivation (comparison group) 135 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Page No. Table Al. Odds Ratios from Multinomial Logistic Regression of Being in Economic Upward Transfer vs. No Upward Transfer (comparison group) 163 Table A2. Odds Ratios from Multinomial Logistic Regression of Being in Economic Upward Transfer vs. No Upward Transfer (comparison group) 165 Table A3. Odds Ratios from Multinomial Logistic Regression of Being in Full Upward Transfer vs. No Upward Transfer (comparison group) 169 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. XI LIST OF FIGURES Figure 1. Four Directions and Types of Intergenerational Transfers Between Two Generations 56 Figure 2. Model Fit Evaluated by BIG for Classes 1-5 118 Figure 3. Conditional Latent Class Probabilities for the Emotional and Instrumental Upward Transfer Class 122 Figure 4. Conditional Latent Class Probabilities for the No Upward Transfer Class 123 Figure 5. Conditional Latent Class Probabilities for the Economic Upward T ransfer Class 123 Figure 6. Conditional Latent Class Probabilities for the Full Upward Transfer Class 124 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. XII ABSTRACT Taiwan is at the forefront of modernization in Asian countries that adhere to Confiician values and teachings. The interplay of modem and traditional forces gives rise to different patterns of intergenerational social support exchanges. This study examines how elderly perceptions of filial norms and other motivations affect intergenerational social support between older adults and their adult children in Taiwan. A total of 9,477 dyads are formed for this analysis by matching the responses from a nationally representative sample of older Taiwanese individuals (N=2,763), who are 65 years and older, to each of their coresident and noncoresident children. Four types of transfers emerge, based upon latent class cluster analysis: emotional and instrumental upward transfer, no upward transfer, economic upward transfer, and full upward transfer. These transfers were then used as the dependent variable, with each class being compared to the no upward transfer class. The predictors for the types of transfers used in this analysis were the following four domains of motivations: need, exchange, resources of power, and ideology. Results from the multinomial logistic regression show that under the need domain, older parents who are unhealthy are likely receivers of economic support. Older parents who have economic power, which includes control over property, economic decisions, and a primary income, will likely receive emotional and instmmental transfers rather than no transfers at all. The main model of interest R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Xlll is the filial norms model under the ideology domain, which shows that older adults with higher levels of normative expectations are more prone to receive instrumental and economic transfers from their children. The blending of modernization and traditional Confiician values are occurring in Taiwan, as older adults who hold both more economic power and higher filial expectations are receivers of emotional and instrumental support. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. CHAPTER ONE Introduction The purpose of this research is to examine the motivations for intergenerational exchange between older adults and their adult children in Taiwan. In order to answer the central questions of this study; Has the traditional perceptions of filial piety changed for older people with the process of modernization? If so, what are the intergenerational exchange patterns that accompany these eurrent perceptions of filial norms? Intergenerational exchange is a fundamental part of the social fabric in society, particularly in Asia because of the historically important role of family care for the elderly. Amidst demographic, economic, and social changes taking place in Taiwan, intergenerational support plays an important mediating role for the well-being of elderly, especially in affecting the design and effectiveness of public programs and policies (Ofstedal et al., 2002; Mason, Lee, & Russo, 2001). During the last 50 years, Taiwan has made advancements in population control and public health initiatives, which have resulted in a sharp increase in the older Taiwanese population with an attenuating support network of children, due to lower fertility and more mral to urban migration. Thus, with the lack of sufficient government programs for helping the elderly and their families, this rapidly growing older population is potentially a burden to adult children who face the reality of balancing caregiving duties and work productivity, with less siblings to help share the burden R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. of care. As in other Asian countries that are experiencing modernization while retaining the traditional practice of filial piety, policymakers in Taiwan are reassessing their public support initiatives to complement, substitute for, or reinforce these changing filial responsibilities (Li, 1994). This research seeks to make both theoretical and empirical contributions to the gerontological literature by identifying the models of motivations for intergenerational exchange in Taiwan. This study will use a nationally representative sample of older Taiwanese individuals, who are 65 years and older, in order to test several models of intergenerational transfers by examining patterns of intergenerational social support and their predictors. Although the literature shows researchers have investigated a variety of motivations for support exchanges in famihes, such as the altruism and power-and-bargaining, there has been scant attention paid to filial piety as an enforcement mechanism promoting various types and levels of such exchanges. Since filial norms are taught through the educational system and passed on from one generation to the next in Taiwanese families, this dissertation will examine intergenerational transfer patterns through the manifestation of these filial normative perceptions among the older Taiwanese, while taking into account other motivators for exchange. This dissertation also seeks to make a methodological contribution by simultaneously examining intergenerational support received and given, rather than looking at each separately, as is found in most research studies (Agree et al., 1998; Biddlecom, Chayovan & Ofstedal, 2002; Lee & Xiao, 1998; Lin et al., 2002). Based R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. upon the empirical literature to date, four main types of exchange patterns are innovatively hypothesized to occur between older and younger generations (Agree et al, 1998, 2002). The first is reciprocal exchange, where transfers flow upward and downward across generations. Upward transfers only, where support flows up from adult child to older parent, characterize the next type. The third type is by downward transfers only, where support flows from the older parent to the adult child. The last category is unique by the absence of transfers where support is neither provided nor received from the adult child. This study will use latent class cluster analysis (LCA) to describe latent classes of intergenerational support exchanges. LCA is potentially an important tool that can be used to formulate typologies that are otherwise undetected by traditional models, such as regression, discriminant, or log-linear analysis containing parameters describing only relationships between the observed variables. Latent class models differ from these by including additional parameters that describe relationships between observed and one or more latent variables. The findings from this dissertation will have the potential to allow policymakers in Taiwan to gain a better understanding of overall expectations for informal support, the types of intergenerational exchanges that actually occur, and the possible motivations for intergenerational transfer patterns; ultimately such information can be used to develop better and more universal formal support programs. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. CHAPTER TWO Background of Taiwan Relative to Other Asian Countries The chapter will begin with a discussion concerning the prevalent social assumption in Taiwan and other Asian countries that families are still able to continue traditional roles of caring for the aged despite the pressures of population aging. The next section will summarize the overall demographic, economic, and social changes of Taiwan in relation to other Asian countries with regard to caring for older people. The last section of this chapter will describe the uniqueness of Taiwan’s geography, history, and culture that researchers will need to understand in order to explain family behaviors within the country. Can Asian Families Continue to Support the Older Generation Despite the Consequences o f Population Aging? As in many other Asian societies, the concept of filial piety is a prevalent social norm in Taiwan. It is based upon Confiician teachings that emphasize the importance of respect and support offered by younger family members toward their elders. The kinship system under filial piety is a social mechanism of stability, allowing a person to expect a secure support system during old age (Chao, 1983). The mutual obligation that characterizes filial piety is the underlying assumption that a firm basis exists for the provision of continuing care for R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. frail elderly parents (Li, 1994). Based upon this assumption, governments in Asia are attempting to avoid costly public investments in institutional care and programs for the aged by holding the family responsible for their older members through public laws. However, with the economic and social changes associated with population aging, the following question arises: Will families actually have the capacity to continue caring for their elderly through traditional means of support without any help from the government? The answer to this question is dependent upon three main factors: socio-economic status, socio-cultural norms, and family structure. The first factor is the socioeconomic circumstance of a family. If the socioeconomic status of a family is high, then providing traditional means of support for its elderly members may still be feasible. However, for families who are living in poverty and are not receiving adequate welfare subsidies from the government, this will put strains upon family members to provide care for multiple generations. As of 2002, only 0.68% (153,337 people or 64,535 households) of the Taiwanese population lived in poverty due to the success claimed by the government of providing for these low-income individuals and families through welfare programs. In addition, for low-income older persons who coreside with their families, there are social welfare subsidies that they can qualify for that are added to the family welfare subsidy. Therefore, the answer to the above question is “yes,” families in Taiwan are still able to provide for their elderly based upon the low poverty rates and availability of social welfare for those who are low-income. However, poverty rates may begin R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. to rise as the Taiwanese economy began to suffer from an economic downturn as of 2001. This was attributed to the September 11, 2001 terrorist attacks in the United States with external trade falling dramatically, not only dampening domestic production, but also leading to a 29.2 percent plummet in real private investment and a higher unemployment rate of 4.6 percent. Taiwan suffered from declining income, rising unemployment rates, stock market declines, and the New Taiwan dollar depreciated by 7.6 percent against the US dollar (GIO, 2003). Thus, with a continuing sluggish economy, there may be an increase of multigenerational families who will struggle to make ends meet, but do not qualify for welfare subsidies. In addition, the elderly in poverty who do not have family to live with are penalized in the sense that they receive government support that is approximately 50% less than their counterparts who coreside with their family (GIO, 2003). The second factor is the socio-cultural norms within which the family functions. Younger Taiwanese family members who are wealthier and able to provide elder care through the traditional practice of coresidence with older generations may choose not to do so. Younger generations are not only displaying shifts in attitudes and values toward independence, but older people may also choose not to follow traditional norms by living independently from their children and grandchildren. This may be because they are in good health and have adequate economic resources (Biddlecom, Chayovan, & Ofstedal, 2002; Thornton et al, 1994). There are also adult children who choose to practice filial piety through nontraditional means, such as sending financial support to their older parents or R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 7 hiring professional caregivers to provide in-home care for their parents, rather than directly providing the care themselves (Lee and Xiao, 1998). Third, as a result of industrialization and urbanization, the structure of the Taiwanese family is shifting from multi-generational to nuclear units (Phillips, 2000). Modernization has had varied effect upon family composition and resources, and younger members are now more likely to migrate to urbanized areas seeking higher education, better salaries, and independence from traditional values and norms. In Taiwan, these changes are already producing severe complications for the traditional norms of family interdependence and exchange of support. This has been compounded by the fact that as of 2000, the birth rate was down 1.4% and the death rate had similarly declined, down to 0.568%, resulting in a larger elderly population with a smaller informal support network of family caregivers (GIO, 2002). With population aging, these demographic changes will produce many challenges along with the familial economic and social transformations that are occurring in Taiwan. Demographic, Economic, and Social Changes in Taiwanese Families in Relation to Other Asian Countries There has also been a growing body of literature based on research related to intergenerational social support in Asian societies (Agree et al., 1998; Chen & Silverstein, 2000; Domingo & Asis, 1995; Freedman, Thornton, & Yang, 1994; Hermalin, 2002; Philips, 2000; Shi, 1993). The underlying theme has been the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. convergence of demographic and socioeconomic changes in Asia has led to concern for the vitality of the society and economy, the strength of intergenerational relations, and the well-being of a rapidly growing older segment of the population. Overall, Taiwanese families have many similarities families in other Asian countries, but there are also differences routed in Taiwan’s unique historical context. Demographic Changes The Asian family is functioning within a virtual pressure cooker due to the stress of population aging occurring across most of the region, in combination with the slowness of public policies to respond. With population aging, particularly due to declines in fertility and high rates of out-migration by younger generations, older parents are facing diminishing social network structures and limited access to resources to ensure their well-being in old age. The government will need to face this challenge of providing care for the elderly alongside the working population, in order to ensure both the continued productivity for the country and the welfare of the elderly. To further understand the problem, this section begins with an examination of the demographic changes occurring in Asia. Fertility and mortahty rates have been dechning in most parts of Asia, the result of urbanization, and rising education levels and age at first marriage. As fertility rates continue to decline, the situation becomes even more serious when taken into consideration with the increased longevity being experienced by all members of society. These factors combine to produce an even greater burden for R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. the younger members of society who will have to care for multiple generations of older family members at the same time. Currently, the majority of the world’s older adults live in Asian societies, with 52% of people aged 60 and over and 39% aged 80 and over residing in this region. An increase of 5% and 9%, respectively, is expected in these rates by 2020 (Randel et al, 1999). Recent projections show that by 2025, 15% of the total world population will be older Asians that are 65 years or older, up from about 9.1% in 1980 (Chang, 1994; Phillips, 2000). According to Kinsella (2000), over a quarter of Japan’s population will be aged 65 or older by 2025. It is for this reason that Japan is struggling with negative population growth, which creates the challenge of finding incentives for women to bear more children to support future older generations. Conversely, countries such as Cambodia and Laos, which have significantly younger populations and are introducing family planning policies, will have only 3.5% and 5.6% of their total population be 65 or older by 2025 (Kinsella, 2000). In between these two extremes lies Taiwan, which is projected to have close to 18% of its population consisting of adults aged 65 and over by 2025 (Lee & Sun, 1995). The varying proportions of older persons in the total population show the heterogeneity of the Asian region, but given their closeness and occasional overlap in geography, history, and culture, the more modemized countries can be studied as a type of forecasting mechanism for other countries in the region that are currently modernizing. As such, Taiwan has been a laboratory for many researchers interested in the aging experience of developing Asian countries undergoing rapid social, R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 10 demographic, and economic transformations during the past 50 years (Bartlett & Wu, 2000). In Taiwan, the population structure has undergone tremendous changes over the last half century. The total fertility rate has dropped from 5.9 children per woman in 1949 to less than 1.38 children per woman in 2000 (GIO, 2002). Mortality levels also declined sharply over this period, with average life expectancy at birth advancing from 51 years in 1949 to 75 years in 2000. The combined fertility and mortality changes have led to an older age structure, with a sharply lower proportion of the population younger than 15 and an accompanying increase in the proportion 60 years of age or older. The 60 and older segment of the population has increased from 4.7% in 1949 to 8.81% of the total population in 2001 (GIO, 2002). As those bom during the baby boom grow older, the economically productive 15-64 year old age group increased to 70.26% of the total population in 2000. While the current ratio of adult children to older parents is currently quite high, when the baby boomers reach older ages there will be a shortage of workers to provide economic support to the older generations within the family given the decline in fertility rates and increase in longevity. As in the United States, there is great concern over the aging of the large amount of post-World War II “baby boomers,” many of whom already have or will soon be joining the ranks of those 65 of age and older. The parallel decline in mortahty after the second World War, particularly among children, also means that a higher proportion of this large cohort will survive to old age. If these retiring R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 11 individuals continue to follow traditional means and expectations of support, this will add yet more strain on younger family members. However, some scholars argue that, as in the United States, the large baby boom cohort may not fare worse than their parents, given their attainment of higher levels of educational and occupational status. Baby boomers are already beginning to take more personal responsibility for their retirement through company pensions and investments, relieving some of the pressure off the next working generation (Chang, 1994; Hermalin, 2002; Lee & Sun, 1995). In summary, the demographic changes and projections for population aging in Asia indicate that there will be rapid increases in the proportion of older individuals in the total population, particularly among the oldest-old (those age 80 and over). With the largest proportion of older adults in the world, the elderly Asian population will continue to grow to three times its 1995 size by 2030, placing tremendous strain upon families, health programs and facilities, social services, and future retirement programs. Phillips (2000) refers to the current cohort of elderly in Asia as an “interim generation” in need of formal support for assurance of care, and argues that elderly Asians without substantial personal resources in rapidly developing economies and living in technologically unfamiliar environments may suffer from poor health and poverty if state and family resources are not available. Some countries, such as Taiwan, are initiating and modifying their basic social welfare programs for the elderly on a local level, while other countries are making major headway in developing these policies at the national level. However, these R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 12 efforts may not be enough, as 20-25% of the elderly population in Asia are already facing a fair degree of poor health and economic hardship, which will translate to a larger number of at-risk older people with the acceleration of population aging (Hermalin, 2002). Economic Changes Amidst the changing demographics in Asia, the overall economic growth experienced in the region is an underlying factor in supporting the increasingly growing older generation, as most of these countries do not have a formal income retirement policy. With flourishing economies, the younger generations have more accessible resources to support the older generations, as they are able to find and retain employment with higher income in urban areas and with increased access to higher education. The countries in this region have experienced overall growth in varying degrees over time. For example, Japan's economic dominance is now waning while China is considered an emerging economic power. There are also internal fluctuations in the economy for the Asian Tiger countries - Taiwan, Hong Kong, Korea, and Singapore - that experienced rapid prosperity prior to 1990 (Phillips, 2000). In comparison to other Asian economies experiencing an average of 4.4% gross domestic product per capita growth rate between 1950 to 1999, Taiwan was experiencing a phenomenal growth of 5.6% during the same period, deeming it an "economic miracle" (MOEA, 2002). The rapid social and R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 13 economic changes that has occurred in Taiwan during the last twenty years has put the country as the 1 4 * ^ largest economy in the world as of 2002 (GIO, 2003). However, as with other countries in the region, Taiwan has experienced varying cycles of economic growth, which has been fluctuating with unemployment falling to the its lowest as of November 2003 at 4.71 percent along with the non-performing loan ratio hitting 5.01 percent from its historic peak of 8.04 in January 2002 (GIO, 2003). Reliance upon the economy is precarious with the fluctuations that occur, which means that a formal income policy for elderly needs to be in place to help the working population support their elderly relatives when the economic growth is slower or negative. Economic growth is often the product of technological and scientific developments that make productivity more efficient and in larger capacities, often shifting the economy from family enterprise to wage employment (Chang, 1994). Consequently, there is also shift in the spheres of authority from the older to the younger generations, as older male members of the family have less control over knowledge, skills, and financial and material resources for productivity in family economies, while younger men and women can obtain alternative means of employment in urban areas, such as in factories or offices. The traditional loyalty and obedience of the younger members, which is closely attached to the control over these resources by the elders, is therefore reduced, resulting in less assurance of support and care promised and given to older adults in the same residence. Further, unless these older adults continue to expand their knowledge, skills, and economic R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 14 resources to allow them to live independently from their children, there are potential consequences of having more elderly living in poverty. As in other Asian countries undergoing industrialization, Taiwan also displays this phenomenon of high industrialization and urbanization. During the past 50 years, higher average per capita national product (GNP) growth from $200 to $14,188 between 1951 and 2000 in eonstant US dollars has rapidly increased the growth of living standards (GIO, 2002; MOEA, 2002). Although higher income levels have allowed adult children to achieve more privacy via separate households in urban areas from their older parents, it has been found that over half of the population aged 65 and over in Taiwan are financially dependent on their children (Comman et al, 1996; Directorate-General of Budget Accounting and Statistics, 1999). In Taiwan, accompanying these economic changes is the expansion of educational opportunities from 1950 when there were only 62 senior high schools serving 18,866 students island-wide to the current 277 schools with 356,589 students (GIO, 2002). During the same period, there was an increase of higher education institutions, which include public and private four-year universities and junior colleges. As of 2000, there were 150 higher education institutions with a total of 1,092,102 undergraduate and graduate students, which is 164 times the number of students who attended only seven institutions that offered higher education programs in 1950 (GIO, 2002). Women graduates from higher education institutions have increased by 50 percent from two decades ago, constituting half of all graduates from universities and colleges and close to one-third of those from graduate schools (GIO, R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 15 2002). This rapid increase in female graduates is paralleled by a rapid rise in female labor force participation, which is approximately 50 percent of the female population in Taiwan. The higher education and labor force participation pursued by women is also associated with the average age of marriage increasing from 25.8 in 1990 to 26.1 in 2000, while for men there is an even greater delay from 26.7 to 30.3 years old (GIO, 2002). This delay in marriage also means that unmarried children are likely spending longer periods in the parental home. The growing labor force participation of married women outside the home may limit the amount of support daughters-in- law can provide their in-laws, even when parents coreside with one or more married sons. As a result, aging parents are now playing an increasingly important role in childcare when the young couples are at work. In a study of four countries in Asia, Biddlecom, Chayovan, and Ofstedal (2002) found that the largest amount of older adults (70%) in Singapore who say they care for their grandchildren, followed by 38% in the Philippines, 32% in Thailand, and 23% in Taiwan. With countries and families facing demographic and economic changes, the family has undergone many structural and functional transformations as well, but normative dimensions of attitudes and values are still relatively intact in Asia, which will be discussed in the next section. Social Changes R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 16 An important factor in supporting the older generation in the region is the widespread and continuing public and private affirmation of the importance of family support. The prevalent value system of Confucianism upholds the family responsibility of caring for the elderly, which underlies most laws in these countries as the basis of having elderly welfare as a private matter. Thus, there is minimal state provision with continuing dependence on informal support, insurance, and savings for care. With this predominance of filial piety under the varied effects of modernization, this section will describe and explain the aspects of family composition and resources that have remained intact and other aspects that have undergone transformation. The traditional and direct manifestation of filial piety in society is coresidence; currently, an estimated 65.7 percent of older Taiwanese live with their adult children (GIO, 2002). Further, studies have found that the majority of older people prefer to live with or nearby their children (Bartlett & Wu, 2000; Knodel & Ofstedal, 2002). In a survey of public attitudes towards long-term care, 56 percent suggested that families take care of their dependent elderly in their own home; 32 percent suggested community care; and 10 percent suggested institutional care (Wu & Chu, 1996). The main reasons older people were not living with their children were due to childlessness, children living overseas, or accommodations were too small, especially in urban areas (Chan, 1992; Li, 1994). The demographic and economic changes described in previous sections seem to have greatly affected the actual living arrangements of the elderly R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 17 to-date. Between 1985 to 1999, Knodel and Ofstedal (2002) estimates that the proportion of those over 65 living with married children declined from 83.7 percent in 1976 to 69 percent in 1996. When compared to other Asian countries in the region, such as the Philippines, Thailand, and Singapore, Taiwanese elderly have experienced the sharpest reduction in coresidence rates during the last two decades. Chen (1998) identifies four types of living arrangements in Taiwan. The first type is the elderly who moves in with one or more adult child to assist and accompany them. The second type is the elderly who moves away from his or her family for personal reasons, such as a health change or moving into a nursing home. The third is marriage dissolution, which will cause an elderly to go from living with a spouse to living alone. The last type is attitude toward living arrangement, which is a significant predictor for whether or not someone will be living alone, with a spouse only, or with his or her family. Chen (1998) shows that a favorable attitude toward living with adult children is observed to negatively affect the probability of living alone. Those who live with spouse only, but who prefer to coreside with their children, will likely transition to living with their spouse and children. Despite the large proportion of intergenerational coresidence that still exists in Asian countries, such as in Taiwan, there also is a trend toward increasing unmarried household heads, especially among young women (Thornton et al, 1994). With increasing educational attainment, income, and urbanization, these unmarried heads of households are earning and spending the most on themselves, while those who are in stem-families show the opposite. These stem families stay in this type of R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 18 living arrangement based on economic necessity that family members need to pool their resources together by spending money on basic items, such as food (Hseuh, 2002). The shift from the traditional stem family to the one-person household for the elderly has been associated with economic and social factors of urbanization, higher education, and migration. Whether for better educational or occupational opportunities, many children have moved away from their parents to urban areas (Chang, 1994). Thus, allowing children to interact with more non-family members, often choosing their own mates, which has traditionally been the decision made by the parents. Thornton and colleagues (1994) point out that the rise in urban migration from rural areas produces a higher concentration of nuclear households in the latter. In addition, those who choose mates who have family living in the cities may form new extended households (Thornton et al., 1994). Furthermore, older parents who had stayed in the rural areas when their children migrated to the cities may later decide that they want to move to the city to live with their children. At the same time, young people who grew up in cities may have less motivation than their rural peers to migrate away from their parents because they can locate jobs in their own communities (Lee & Sun, 1995). Because of these migration dynamics, it is not surprising that there is an increasing shift from nuclear to stem families in Taiwan during the recent decades. Another direct manifestation of migration is the increase in children who look after the physical and economic well-being of parents living in separate R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 19 households, but in the same neighborhood (Hermalin, 2002). There are also indirect ways of showing filial piety, particularly for adult children living in other towns or cities in Taiwan or overseas, by sending money to their older parents or hiring paid help to provide instrumental care (Agree et al., 1998). However, there are more adult children who have become cautious about unquestionably obeying their elderly parents because their educational attainment of knowledge surpasses that of their older parents. Despite of this status differential, parents still have expectations of their children. In a nationally representative longitudinal survey in Taiwan, there were four of ten prevalent expressions of filial norms that older adults still expect from their children measured, between 1989 to 1999 (Biddlecom, Chayovan, & Ofstedal, 2002). The first is “children should consult parents about major decisions,” the level of participation for older persons as a family decision maker is positively related to family size (Williams, Lin, & Mehta, 1994). Next, is “children should provide instrumental support.” The majority of older adults (80%) still feel that they should receive support from their adult children. “Children should provide money” is the third expression with the majority of older adults expecting partial, not full, monetary support. Lastly, between 1989-1999 elderly Taiwanese have held positive and stable views that their “children will provide support overtime” (Comman et al, 2001). The demographic, economic, and social changes found in Taiwan are similar in many respects to countries that are either undergoing or have undergone R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 2 0 modernization, such as China and Japan, respectively. Throughout this section, the prevalent theme is that these changes are producing pressures upon family members who are obligated either by law or normative values to continue carrying out their traditional roles of caregivers for older generations. Researchers and policymakers will need to continue examining the long-term effects of these transformations upon the family in finding the balance between formal and informal support for the aged. Many scholars have argued that the elderly in Asia are facing an emerging crisis of substantial deficiencies in personal resources and living in unfamiliar technological environments (Chang, 1994; Kinsella, 2000; Phillips, 2000). Hence, older persons under these circumstances are deemed by Phillips (2000) as the “interim generation” of older persons, arguing that if state and family resources are not available these older persons will suffer from poor health and poverty. Uniqueness of Taiwan’ s Geographical, Historical, and Cultural Impact upon Families As seen in the previous section, Taiwan has many similar characteristics with other Asian countries. However, Taiwan has its own unique geography, history, and culture that are indispensable for understanding ethnic differences in the behaviors of family members and the ethnic differences within and outside of the country, particularly with China. Although Taiwan is located only 100 miles off the coast of China, the tropical island was first discovered by the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 21 Portuguese in 1544, naming and mapping it as Ilha Formosa meaning “Beautiful Island” for its lush vegetation, forested mountain ranges, and beaches. This discovery by the Portuguese sets the historical tone for the series of foreign influences upon Taiwan leading forth to its path toward modemization. Due to its unique geographical and historical streams, the culture of Taiwan has evolved in such a way as to affect the social development of the family. Geography Located on the Pacific Rim in the Taiwan Strait is the island of Taiwan that has a total area of 36,000 sq. km., equivalent to the size of Delaware and Rhode Island in the United States. One of the largest and oldest industries on the island is fishing, which has grown from the small-scale coastal fishing by a hundred family trawlers to deep-sea commercial fishing by 26,623 ships, but this accounts for only 0.07 percent of the current total workforce in Taiwan (GIO, 2002). Since Taiwan is mostly made up of mountains, the agricultural industry is limited to the island’s arable western slope lands and alluvial plains. Farm plots tend to be small with 76 percent of all farming households having less than one hectare of arable land, with farmers currently making up only 7.4 percent of the total workforce (GIO, 2002). Although there have been efforts made by the government to redevelop the oldest industry of agriculture toward sustainable development, the limited land available for farming has led the island toward the path of scientific and technological development. Prior to 1990, Taiwan’s economic achievements were R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 22 primarily based on labor-intensive industries and low-tech export products. In the past decade, 54 percent of Taiwan’s exports were high-tech products and only 14 percent were low-tech, making the country the world’s fourth largest IT hardware producer (GIO, 2002). The industrial production sector consists of 27 percent of the workforce as of 2002 with the rest of the majority (65.53%) in the service sector, which ranges from commerce to governmental services, similar to other modernized countries (GIO, 2002). Hence, due to the limited size of the island, the growing population has inevitably caused the family economy to shift from agricultural and fishing to the service and industrial sector. History When the Portuguese landed in Taiwan after the mid-1 6 * * ^ century, they found large tribes of indigenous peoples from proto-Malayan ancestry, who’s descendents currently make up two percent of the ethnic composition in Taiwan. Earliest records show that the Portuguese left the island after staying for only six weeks on the southwest coast (Davidson, 1903). Then the Dutch East India Company established a military base in southern Taiwan in 1622 and ruled the island for 40 years fighting against the Chinese and the Spanish, who occupied the north for 16 years. The Dutch were instrumental in developing mercantilism, agriculture, and schools in Taiwan (Campbell, 1903). They also induced the Fukienese from the Fujien province in China to migrate to Taiwan growing sugarcane and rice in the 1630’s, which began the first large-scale immigration wave that make up the current largest R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 23 ethnic group (60%), known as the Holo (Meskill, 1979). An agricultural revolution began with the Holo working under the European tenancy system, which allowed the Holo to live on the land by producing crops and paying heavy rents to the Dutch. Due to immigration, the Holo tended to have fragmented lineage groups and forged new kinship and marital ties with other immigrants, indigenous peoples, and even with the colonizers (Olds, 2003). Thirty years later, with resistance wars in southern China occurring, another group of immigrants called the Hakka came from the eastern Guangdong province, and were forced to settle in the hilly and more mountainous parts of Taiwan because of the friction with the Holo group for industrious farming land. When compared to the Holo, the Hakka maintained strong solidarity and organization of their family lineage and currently make up about 25% of the population in Taiwan (see Lamely, 1981). In 1662, there were approximately 40,000 Holo and Hakka people living in Taiwan (GIO, 2002). It was in 1662, when the Dutch were defeated by the Chinese pirates led by Cheng Cheng-kung, also known as Koxinga, who ruled Taiwan for 22 years with his descendents before surrendering control of the island to the Manchus in 1683 (TAAGP, 2003). A different ethnic group from the Han Chinese, the Manchus were the last dynasty to rule China, known as the Ching. While agriculture was expanding across the island and exports were made to China, Japan, and Australia, foreign occupation persisted with Japan occupying southem Taiwan in 1874 and the French attacking northern Taiwan in 1884 until the Ching finally realized the importance of Taiwan as a gateway to the seven provinces along China’s southeastern coast and R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 2 4 officially annexed the island as a province for eight years beginning in 1885 (Bing, 1986). Then in 1895, Taiwan became a Japanese colony, after China lost the Sino- Japanese War. The Japanese development of Taiwan included railroads, agricultural research, public health, banking, education and literacy, and business. Following World War 1 1 , Japan granted Taiwan sovereignty in 1945, but the Allied forces retroceded the island to the Republic of China (ROC), under the Kuomingtung (KMT) party leader—Chiang Kai-Shek (Kerr, 1974). It was in 1947, after Chiang Kai-Shek lost the Chinese Civil War to the Communist party, that an influx of an estimated one and a half million soldiers and civilians fled the Communist People’s Republic of China, leaving their families behind, to the relocated Republic of China (ROC) on Taiwan, beginning the authoritarian regime under Chiang and the cold war that is still in existence (Kerr, 1974). The Chinese immigrants were government officials \ business leaders, and large number of KMT military personnels (Thornton et al., 1994). While the two former groups were well educated and most likely came with their family members, the latter were young single men with limited credentials, who represented a significant proportion of the young adult population. This produced a larger discrepancy in the male to female ratio, and in turn, more competition between the Chinese, Holo, and Hakka men to find spouses. This caused many of the Chinese immigrant men to either delay marriage or remain ^ At the time, these men made up only 3% o f the population, but were allocated 97% o f the government positions, which produced unfair representation for the rest o f the people in Taiwan. But under the KMT authoritarian regime and biased election processes, this misrepresentation occurred for over fifty years, until 2000, when fair democratic elections finally took place in Taiwan, peacefully handing over the reign o f the KMT to the Democratic Progressive Party (DPP) o f Taiwan. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. single, and those who did marry found spouses who were, on average, 10-20 years 25 younger than they were (Thornton, 1994). This large age gap would influence the basic demographic character of Taiwanese society producing a larger number of older widowed women who married these retired soldiers. The aging of this cohort of Chinese comprises an important component of the elderly, and because of their distinct history and pattems of family formation, their exchange pattems are of special interest in research. Culture Each of these ethnic groups have one obvious similarity, which is the importance placed on caring for the elderly. However, with their own unique geographic, historical, and cultural experiences, each group will also have unique manifestations of how they care for the aged. The principles which family authority was patterned after emphasizes the dimension of generation, age, and sex (Greenhalgh, 1985). In the Chinese society, the oldest male ultimately defined decision-making authority over family issues, economic cooperation, and inheritance, in the oldest living generation. The power of the older over the younger generation lies in the adherence to "filial piety," which stresses the absolute obedience of a son to his father throughout their lives. Even upon death, the oldest son carries the daily responsibility of preparing tea and buming incense for his parents and is ancestors (Chao, 1983; Lee & Sun, 1995). The oldest son’s wife carries out these rituals, in reality, as daughters become a member of her husband’s R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. family upon marriage. Thus, children were socialized and educated to center their 26 activities surrounding the family as authorized by the senior male figure. As Taiwan has experienced increased secularization of beliefs and worldviews with modemization and scientific developments, there has been tremendous change in its marital pattems. There has been an increase in the average age at first marriage for both males and females in Taiwan, most noticeably among women from 21.1 in 1965 to 26.4 years old in 2002 (GIO, 2003; Lin, Lee & Thomton, 1994). Among 20-34 year olds, the percentage of women who are married has declined from about 79% in 1965 to only 45.6% in 2000 (GIO, 2002; Thomton et al, 1994). Concurrently, divorce rates in Taiwan more than quadmpled between 1970 and 1990 across all ages, with the highest divorce rate among men and women 20-44 years of age (Lee & Sun, 1995; GIO, 2003). Hermalin (2002) found higher rates of divorce and delayed marriages and lower numbers of children among Taiwanese Baby boomers compared to their parents, similar to what Easterlin and colleagues (1993) found for baby boomers in the United States. In Taiwan, these trends may lead to disadvantages in social support for aging boomers when compared to their parents. Combined with increasing longevity, over time these marital and divorce trends will lead to shifts in the marital status composition of the children providing support for the elderly, with more single children providing support for their older parents. Overall, based upon the similar development of demographic, economic, and social characteristics of Taiwan with other Asian countries, it is likely that Taiwan in the next 15 years will become more like Japan as one of the fastest growing R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 27 populations of the oldest-old (Phillips, 2000). Other developing countries may soon follow the path of modemization and undergo similar transformations as Taiwan. Specifically, China’s economic development, especially the IT industry, which currently ranks third in global standings already surpassing Taiwan, and the one- child policy in major urban areas are becoming areas of research interest for examining the social support for older Chinese. Hence, Taiwan serves as an important country for researchers interested in examining the motivations behind family members caring for the aged against the backdrop of having little or no pubhc policies and programs to support these caregivers and the older generations. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 28 CHAPTER THREE Theoretical Background and Literature Review In this chapter, the definition and importance of studying intergenerational transfers will be established for this dissertation. Next, a review of theoretical perspectives and studies related to the pattems of intergenerational transfers that occur in Taiwanese families will be given. Finally, the theoretical framework for this dissertation will be outlined, followed by the specifications of research questions and hypotheses. What are Intergenerational Transfers? And why are they Important? Intergenerational transfers are the economic, emotional, and instmmental support given and received between generations within a family. These transfers of support between generations are a fundamental social fabric in society, particularly in Asia because of the historically important role of family care for the elderly and the children. Scholars researching intergenerational ties have found that transfers of money, property, time, and goods are not constant, but rather dynamic in nature (Altonji, et al., 1992; Cox & Rank, 1992; and McGarry & Schoeni, 1997). The basis for transferring these resources within families is in line with the needs and capabilities of the generations as they pass through the life cycle. But how does R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 29 intergenerational redistribution take place when there are more needs of nonworking individuals, such as the elderly and the children, then there are resources provided by the working adults? Researchers and policymakers are forced to face this question in a modernizing country as they develop public programs, policies, and studies to provide help to the nuclear family, which is no longer a constant for providing care to the children and the elderly in the public welfare equation. Despite its complexity and the changing nature of intergenerational transfers, scholars have noted the importance of examining these intergenerational pattems in families as they play a cmcial mediating role on the well-being of elderly (Biddlecom, Chayovan, & Ofstedal, 2002; Mason, Lee, & Russo, 2001). In almost all cultures, the central intergenerational tie is between parents and their children (Phillips, 2000). But, as noted by Silverstein & Bengtson (1997), it is a near universal fact that family support can take on a variety of pattems and meanings in different social contexts. And whatever the cultural prescripts, actual pattems of redistribution between family members are further influenced by prevailing social and economic conditions, and can vary significantly depending on individual circumstances and actions. Hashimoto and Kendig (1992) use the concept of reciprocity as a starting point for understanding the structure of family support. One can choose or feel obligated to provide resources to a particular person at some particular time or under certain circumstances. Those who have received support can have reciprocal feelings and obligations, which in tum may lead them to provide when circumstances change (Cox, 1987). The nature of support and the binding R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 3 0 forces, which hold the generations together are multidimensional. These can include affection, obligation, financial power, threats of social disapproval, and other factors in complex combinations. The kind of support—financial, instrumental, or emotional—also may vary from one social context to another. Intergenerational Transfers in Taiwan There has been a growing body of literature on intergenerational social support in Taiwan (Agree et al., 1998; Freedman, Thomton, & Yang, 1994; Hermalin, 2002; Phillips, 2000; Shi, 1993). As presented in the previous chapter, the Taiwanese family is at a crossroads of adapting to demographic, economic, and social changes. These key macro-social changes of lowered fertility, longer life expectancies, increased female labor force participation, and a shift from agriculture- based economies to manufacturing and service-based economies are competing with the demands of filial obligation that form the traditional foundation of intergenerational social support systems (Biddlecom, Chayovan, & Ofstedal, 2002; GIO, 2003). How will families in Taiwan adapt and balance these seemingly competitive macro- and micro-social forces? Why are these traditional norms of obligations still strongly adhered to in a rapidly modemizing Taiwan? This section will be divided into two sections, beginning with a review of the evolution of the modemization theory, along with the empirical evidence that have been lacking in prior studies, as related to Taiwanese families. Next, micro-social level theories and studies specifically related to the pattems of intergenerational transfers that occur in R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 31 Taiwanese families will be given, followed by a discussion of the integration of these theoretical perspectives to provide a framework for this dissertation. The Influence of Modemization on Intergenerational Transfers in Taiwan What is modemization? It is a continual dynamic interaction of social, economic, and political systems as human beings increasingly take control over the effects of industrialization, migration, and urbanization. As predicted by Karl Marx, one of the earliest modemization theorists, each society is progressively moving toward modemization, with industrialization as a key element that produces persistent social and cultural consequences, such as rising educational levels and resistance towards authority (Inglehart & Baker, 2000). One of these consequences, as Cowgill and Holmes (1972) and Pahnore and Manton (1974) argues is that the status of older people declines with modemization, as younger generations embrace values of independence upon migrating to urban areas that offer higher educational and occupational opportunities; thus, abandoning their older parents in rural areas with obsolete skills, minimal resources, and less decision-making power in the family. Such a bleak depiction of the future of elderly under modemization is unrealistic as found in American and European societies, that the elderly have not only maintained survival because of the development of formal resources with modemization, but they have also continued interaction with family members despite R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 32 distance. At the same time, many scholars have discovered that modemization does not necessarily mean Westernization of cultural values and abandonment of traditional norms as Marx and others have both broadly claimed under the modemization theory (Inglehart & Baker, 2001; Kohli, 1999; Silverstein, Bengtson, & Litwak, 2002). In newly developed countries, such as Taiwan, the effects of modemization seem to show that there are still persistent filial expectations among older adults, but the fulfillment of those expectations by the next generation is in less traditional forms. For example, adult children continue to provide upward transfers, especially in monetary support, to their older parents, but no longer live in the same household with their older parents, in order to attain greater privacy (Biddlecom, Chayovan, & Ofstedal, 2002; Knodel & Ofstedal, 2002). Further, any weakening in filial obligation toward the elderly may be offset by economic growth that enhances the ability of these nations to institute or expand formal mechanisms of care for the older population. Even in developed countries, such as the United States, there is mixed support for the theory. For example, even though formal systems of care exist in the United States, continued emotional support from kin is still important in the lives of older people (Bengtson, Giarusson, Silverstein, & Wang, 2000). In a study of six developing countries, Bengtson and colleagues (1975) found that individual exposure to modemizing experiences produces more positive attitudes toward aging and the aged. Thus, researchers have found the modemization theory to be too broad and unfalsifiable, hence lacking in empirical backing, and also failing to take into account the varying R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 33 historical and cultural contexts of countries faced by older people (Bell, 1973; Silverstein, Bengtson, & Litwak, 2003; Inglehart 1977,1990, 1997). The modemization theory fails to explain and predict how industrialization, urbanization, and technological advancements result in a status inversion, declining resources, and decreasing privileges for the elderly. Thus, the modemization theory has been argued as a better representation of a theoretical framework by which authors can conduct research, rather than used as a theory to explain and predict societal phenomena (Silverstein, Bengtson, & Litwak, 2003). For the purposes of this dissertation, the following names are given to the modifications of the modemization theory: the revised modemization theory by Inglehart and Baker (2001) and the hyper-modemization theory by Kohli (1999). Along with these two theories, this dissertation will also look at what Silverstein, Bengtson, and Litwak (2003) calls the task-specific theory that provides a framework by which to examine and predict potential changes that will occur in developing formal and informal resources for older people in Taiwan. Revised Modernization Theory. By examining empirical data from 65 societies and 75 percent of the world’s population in 1995, Inglehart and Baker (2001) performed cross-cultural comparisons of values found in these countries to determine their level of modemization, while taking into consideration their economic and demographic R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 34 backgrounds. Based on shifts in absolute norms and values between survival and self-expression values and between traditional and seeular-rational orientations toward authority, a country was deemed as a traditional or advanced society. On the traditional vs. secular-rational orientation poles, traditional societies tend to emphasize the importance of pleasing parents at all costs, unquestionable obedience toward authority, and national pride. As for the survival vs. self-expression values, traditional societies tend to emphasize survival values giving priority to economic and physical security over self-expression (Inglehart & Baker, 2001). Thus traditional societies tend to refrain from the following three activities: describing oneself as being happy, signing a petition, and trusting others and foreigners. Countries that are more modernized tend to be more forthright about expressing their own opinions and thoughts conceming issues that are related to political authority. As one of the societies in the Inglehart and Baker (2001) study, Taiwan is considered a modernized country with persistent Confucian traditions. Consistent with other modernized countries is its eeonomic development with a GNP per capita over $10,000 and a shift from agrarian to service and information-based society. In terms of the orientation of values, Taiwan is also considered to be similar to other advanced industrial countries because it is more secular-rational than traditional, but tends to be more survival rather than the self-expressive dimension. The authors attribute the persistence of Confucianism as a reason for the secular- rational and the survival orientation. However, the authors do not explain why the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 35 persistence of the traditional value systems of Confucianism does not place Taiwan and other Asian countries in the traditional dimension of the traditional/secular- rational values orientation, especially since the importance of the family is dominant in Confucian families. According to Zhang (2003), Confucianism embodies similar values as democracy, found in many modernized and advanced industrial nations, with a “belief in hard work, thrift, honesty, self-discipline, high regard for education, respect for enterprise, and concern over family stability.” In particular, there have been misinterpretations by politically hungry male scholars over the original teachings of Confucius, which does not command blind obedience to authority and does encourage accessibility of education to the masses. Thus, the increasing pursuit of higher educational attainment by women within the past two decades has helped Taiwan progress toward modernization as the number of Taiwanese women who graduation from a university or college increased from 36% to 50% (GIO, 2002). In addition, half of Taiwan’s women are regular wage earners and help support their families (GIO, 2002). In terms of the changes in the family structure, Taiwan does indeed display more secular-rational orientations with the shift from extended famihes to over half of the population residing in nuclear families with the last half-century (GIO 2002). At about the same transitional period in 1956, efforts toward family planning began with the development of a social acceptance for abortion, contraceptive use, and sterilization (Freedman et al, 1994). On a broader societal R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 3 6 level, Taiwan’s modernization is also correlated with lower fertility rates during this period (Freedman et al, 1994). Although the persistence of Confucian values has predetermined the modemization process of Taiwan in the secular-rational dimension, the anomaly lies in that the country displays more survival rather than self-expression values inconsistent with other modernized countries. Compared with the rest of the societies in the study, Taiwan maintains medium levels of interpersonal trust and an emphasis upon materialism. However, Taiwan is not completely on the survival pole extreme as the people have had a traditionally high tolerance for foreigners and minorities, due to the historical control of the island by foreigners and minority groups, the results of the 2000 Presidential and legislative elections shows the movement towards self-expression. The Presidential election in 2000 marked a peaceful transfer of power for the first time in Taiwan, as citizens voted towards overturning 50 years of KMT rule by electing a native Taiwanese-speaking President and despite the threat of missile attacks from China. Later that year, the legislative elections also marked a turning point in history when the Democratic Progressive Party (mostly made up of Taiwanese-speaking candidates) won close to 80% of the seats, displaying a shift toward having lower tolerance of foreign (Chinese) authoritarian rule. The results from this study show that there are cultural heritage guides the changes in political and socioeconomic systems under modemization. The work by Inglehart and Baker (2001) and Zhang (2003) shed new light on the anomaly that R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 37 Confucian ideals are consistent with more secular-rational values with similar values found in capitalistic societies, ultimately showing that though there is less coresidence between younger and older generations because of the economic independence of children from their parents. However, under Confucianism, having a secular-rational view does not relieve one of the responsibility to provide care for one’s parents even though it will be in non-traditional ways under modemization. Despite its high correlations within each dimensions of traditional/secular (r=.95) and survival/self-expression (r=.96) in the World Values Survey, the authors have yet to measure its reliability and validity as a comparable cross-cultural measure of values. In addition, although this perspective of the modemization theory considers the contextual factors in various societies and provides an overall backdrop to which researchers can examine changes in society, it does not offer a clear framework regarding the roles and functions that conflict or complement each other by the government and family, which is found in the task- specific theory described below. Task-specific Theory To complement the revised modemization theory, is the task-specific theory, which Silverstein, Bengtson, and Litwak (2003) propose that with modemization there will be an increasing evidence of deficiencies in the family to provide care for its members to keep up with the changing economy, leading to the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 38 development of more government programs carried out by bureaucracies to provide welfare. The family and govemment have seemingly conflicting, yet complementary structures and fiinctions as the govemment bureaucracy maintains technical expertise and impersonal standards for society with the objective of achieving efficiency and economic balance, if not profit; while the kinship group performs the “idiosyncratic tasks” which are motivated by affection and commitment between its members. The task-specific theory helps explain the differences between the public and private transfer patterns for the elderly in rural and urban areas of Taiwan. In a national study by Biddlecom and colleagues (2001) on the introduction of Taiwan’s the national pension program for farmers and the local city and county subsidy programs, they found that there was a greater switch from private to public transfers for farmers than in the local, more urbanized areas. Thus, showing that there are perhaps greater burdens for children with parents in rural farming communities, which is relieved with the implementation of this national pension program because there is a need. At the same time, the local programs do not substitute for the tasks already assumed by the children to support their parents in more urbanized areas. The task-specific theory informs the researcher and policymaker what adjustments requirements of modem families to cope with the contradictions and inherent cross-pressures. On the macro-level, the theory predicts which formal organization (group structure) is most effective for which task that the kinship system lacks due to technological change that deter families from performing functional exchanges. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 39 There are some criticisms about the task-specific theory as it oversimplifies the dynamics of economic, demographic, and cultural factors that are involved in matching the functions to the formal and informal structures (Silverstein, Bengtson, & Litwak, 2003). However, it provides a framework to examine the advancements in developing welfare programs for elderly that can complement the existing services provided by families, which can be further integrated into the revised modemization theory to include the shifts in cultural values that once put the primary responsibility of caring for the elderly on family members to the state. For example, reduced fertility and increased dependency ratios for developing nations create tension in private and public trade-off for elder care at several levels (i.e. adult children in smaller families face greater challenges while smaller cohorts of working adults as well). Yet, prevailing societal values organize social support for dependent members. So even though Western nations are reluctant to interfere with the private sector, European and socialistic countries design public policies to help family. Taiwan, as in other Asian nations, is found to have social values in between Westem and socialistic countries, with filial piety as the norm and the assumption that the majority of older people will be taken care of by their families, social policies are slowly meeting the needs of the elderly in rapidly changing nations with declining multigenerational households. The theory offers an explanation for the motivation behind families to offer care to one another through affection and commitments, which is opposite the objective and technical goals that govemment bureaucracy’s hold to provide care for to society. To further complement the revised R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. modemization and task-specific theories, is another modification of the 40 modemization theory that offers an explanation for the patterns and other motivations for intergenerational support under modemization. Hyper-modem ization Theory Kohli (1999) offers an extension of modemization theory that goes beyond the explanation for development of public and private partnerships under modemization, but that the development of nuclear families and public old age security programs are parallel and mutually reinforcing processes with the state eventually replacing family transfers. In the post-industrialization era, the intergenerational family transfers in Westem nations shows evidence for a net downward flow of private resources, mainly material transfers, from the older generations to their adult offspring, i.e. counter to the public transfers. As for instmmental support—for example, help with household tasks (exclusive of caregiving)—the flow of support is balanced between respondents and their children, while for children and their parents-in-law there is a net upward flow. Therefore, the pattems of transfer on the level of the family would move from a net upward flow for less modemized countries to reciprocal exchange to downward exchange to none for most modemized countries. Kohli (1999) talks about two kinds of motivations for family transfers to occur. First, there are inter vivos transfers from one living family member to another, which are either “voluntary” or “capitalistic” bequests. “Voluntary” R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 41 transfers include a range of models from altruism to self-interest, with the former most prevalent in intergenerational relations based on need and the latter as control maintenance. Thus, the targeting of inter vivos transfers to those children who are in a poorer economic position is strongly confirmed for France and the US as well. The family member directs the “capitalist” or entrepreneurial transfers toward the single goal of accumulating as much wealth as possible. Second, there are bequests made from deceased family members to the living, which can be either voluntary or accidental. Just as the voluntary inter vivos transfers are based upon a range of motivations, these types of bequests are similar except that they are promised upon death of the older person. In the US, a large majority of parents during their lifetime give unequal amounts to their different children but divide their bequests equally (McGarry, 1997). “Accidental” bequests occur because the inability for one to predict one’s time of death and are left with savings (Kohli, 1999). In both the US and Germany, bequests and inheritances make up three times the inter vivos transfers. Kohli (1999) points out that part of these bequests and inheritances to the younger generation are made up of public pensions. The logic is that adult children expect that the old age pension can be transferred to them as material support in times of need and as bequests at the times of parents’ death. In Germany, for example, the strongest effects that Kohli (1999) has found for engaging in transfers are the material resources of the parents and those of the labor force status of the children. The older person’s availability of wealth also has a positive effect on R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 42 downward transfers. In addition, the number of adult children is inversely related to transfer giving. In the Taiwan setting, older individuals who are wealthier can use voluntary inter vivos transfers and bequests at various strategies to obtain instrumental support from their children by promising some assets and property before their death, especially to the children who are most in need of money (Lee & Xiao, 1998). Also, there are wealthier, healthier, and older individuals who value their independence and will instead, give money and instrumental care to their adult children, particularly those who are working and have younger children who need babysitting (Biddlecom, Chayovan, & Ofstedal, 2002). This dissertation will look at the motives behind the inter vivos and bequests (i.e., elderly welfare programs) transfers in Taiwan, with hopes of providing insight into their ratio in Asian nations that are becoming more developed and some more Westernized. The Process and Paths of Transformation Toward Modernization in Taiwan A modified modemization framework based on these macro-social theories mentioned above can help explain the relationship between varying degrees of modernization with a country’s shifts in values, formal program development, and pattems of intergenerational transfers for older persons and their children. Countries that are more modemized tend to show more secular-oriented values and self- expression attitudes (revised modemization theory), which will lead to more R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 43 development of formal programs and systems of care for the elderly (task-specific theory) allowing older parents to give more to their children (hyper-modemization theory). As mentioned in the article by Inglehart and Baker (2001), in comparison to the values in more modernized countries, Taiwan is considered as a country that is in the process of modernizing with values that are shifting from survival to self- expression; however, in terms of the secular-oriented and traditional poles, Taiwan is already a secular-oriented society similar to other modemized countries. However, the formal programs providing for the elderly are not as advanced as modemized countries, thus more of the care lies in the informal network, which mainly constitutes the adult children providing financially or instmmentally. Using the task-specific theory to examine the division of labor between the partnership of public and private organization under modernization, Taiwanese families have experienced both a mixture of social and economic transformations within the confines of tradition. As noted by Fricke et al. (1994b), the general persistence of the most successful technological and economic changes are based on past pattems of familial organization. This is specifically seen in labor organizations that parallel family modes of male authority and in passing on the business enterprise to the oldest son. Along with two theories mentioned thus far, the hyper-modemization theory is complementary because it allows one to examine also the direction of transfers under modemization. The result of more females attaining higher education is the trend toward pursuing non-family related work activities. Premarital R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 4 work experience outside the home has been transformed from an activity engaged in by about a third of the women in he earliest marriage cohort to an activity engaged in by over 90% of the women of recent cohorts. Moreover, the experience of nonfamilial work has spread to a majority of women of every background from rural to urban, and every educational level. Similarly, women’s employers today are overw'helmingly more likely to be unrelated to them than in the past at only slightly lower levels than work experience outside the home. Although half of Taiwan’s women are regular wage earners outside of the home, they still continue supporting their families, which can consist of their biological parents, in-laws, spouses, and children. Interestingly, studies have shown that higher wage single women tend to live with their own parents. Kuodel and Ofstedal (2002) have shown that 70% of working single adult daughters coreside with one or more elderly parent. As for married women working outside the home, there has been a fivefold increase across the cohorts between the period of first-time marriage and the birth of the first child. For the period between the birth of a first and second child there has been fourfold increase in mothers working outside the home. With more mothers working outside of the home and living with their own parents, there are more downward transfers of babysitting services and instrumental care given by the older parent. The macro-social changes that have occurred in Taiwan pose several challenges to the govemment that is evident in other countries that do not want to undermine the traditional family arrangements. As mentioned before, Taiwan is a laboratory for Asian countries that are testing whether "modemization" inevitably R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 45 means "Westernization" in terms of social welfare programs and their costs, or whether the govemment can develop novel arrangements that better suit their cultures, resources and needs of the older population. Although it is a well-documented phenomenon in gerontology, that the overall direction and magnitude of informal social support for an older person is inversely proportional to increasing levels of modemization (Cowgill, 1969,1974; Inglehart & Baker, 2001; Kohli, 1999; Silverstein, Bengtson, & Litwak, 2002), this conclusion is mostly based on the experiences in the United States and Europe, which seems contradictory to the experience in the East and Southeast Asia. In both the East and West, there are parallel demographic and socioeconomic forces at work; with the rise of nuclear families and smaller support networks for an increasingly older population, along with children gaining more education and pursuing higher paying jobs in urban locations, the likelihood of older persons losing their traditional means of instrumental, economic, and emotional support are higher (Goode, 1963; Hermalin & Myers, 2002; Kinsella & Velkoff, 2001). Yet, improved socioeconomic status of children can also translate to increased support for older people, insofar as children remain observant of traditional norms and attentive to the needs of their parents (Chang, 1994; Hermalin & Myers, 2002). Thus, the manifestation of modemization upon intergenerational support in Asia is guided by the prevailing attitudes and cultural values, which leads to more efficient development of the partnership between formal and informal support for the care of the elderly. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 6 Micro-Social Level Theories o f Social Support Exchanges in Taiwan With the examination of macro-social theories as the backdrop for this dissertation, this section will be a review of the basic theoretical perspectives for motivations behind intergenerational exchange within the family unit. Several theoretical perspectives have been advanced to examine the types and directions of exchange between generations (Hermalin, et al, 1990; Hogan, Eggebeen, & Clogg, 1993), the characteristics and selection process of the family members involved (Cooney & Uhlenberg, 1992), and the anticipation for intergenerational transfers (Hofferth, et al, 1994; Krause, 1993; McGarry & Schoeni, 1997; Silverstein, Parrott, & Bengtson, 1995). Intergenerational exchanges usually occur within the structures of reciprocal dependence, in which family members are dependent on each other for desired outcomes (Clogg & Goodman, 1985; Eggebeen, 1992; Hogan, Eggebeen, & Clogg, 1993; Kotlikoff & Summers, 1981; Silverstein & Bengtson, 1994). From the sociological perspective, the basis for studying interpersonal exchange networks is most closely related to the social support exchange theory, which explains the motivations of actors to become involved in exchange is to obtain more of the outcomes that they value and the power to control their resources over time (Eggebeen & Hogan, 1998; Molm, Takahashi, & Peterson, 2000). Homans (1961) first formulated the theory based on dyadic exchanges as the building blocks of collective exchange relations, other schools of thought have developed that focus more broadly on exchange networks as systems that are greater than the sum of their R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 47 dyadic parts (Agree et al., 1998). Researchers have used the social exchange theory to empirically examine various populations of families and units of analysis, ranging from dyads of Americans to social networks of Taiwanese, which can include friends and neighbors (Agree et al., 1998; Eggebeen & Hogan, 1998; Fricke et al., 1994). These social networks can include family support combined with, or serve as alternatives to, other informal support provided by neighbors, friends, or fellow villagers, as well as formal support provided through the marketplace, govemments, and voluntary organizations. Fricke and colleagues (1994) describes the family as a “self-sufficient unit” held together by reciprocal ties and redistributing significant resources either at one point in time or simultaneously between sets of donors and recipients. The kinds and amounts of support provided can also vary within as well as between these various types of families. From the nuclear family, made up of a couple who lives with dependent children, to the traditional extended family, in which several generations of related adults and their children interact and reciprocate in coresidence. According to the various forms of the social exchange theory, transfers are predicated upon the expectation of reciprocity (Douglas, 1990). The nature of this reciprocity may be direct (from the individual who receives the transfer) or indirect (through norms that dictate giving within a social group). This review will examine the intergenerational transfers within the Taiwanese family in an expanded framework that includes five main intergenerational transfer models found in the literature. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 48 Intergenerational Transfer Models There are five microeconomic models of intergenerational transfers found in the literature on Taiwan: the altruism, self-interest, power and bargaining, corporate, and the mutual aid models (Lee, Parish, & Willis, 1994; Lee and Xiao, 1998; Lin et al., 2003). These models were developed specifically to explain the transfers of support between generations based on the characteristics of the older parents in Asia. According to Lee and Xiao (1998), the models of motivation are not exclusive of each other, which is dependent upon the characteristics and strategies of the family members involved in obtaining the support that perceive to need. Altruism model. Under the altruism model, adult children give to their parents based upon the perceived need of the parent, which is typically dependant upon their sociodemographic background, economic resources, health status, and use of elderly welfare services (Becker, 1991; Riley, 1983). In both Westem and Asian nations, parents who are older, single, female, low socioeconomic status, living in rural areas, and not receiving any income are likely to be in need of support (Hermalin, Ofstedal, & Lee, 1992; Kao & Stuifbergen, 1999; Lee, Parish, & Willis, 1994; Lee & Xiao, 1998; Shyu, Archbold, & Imle, 1998; Yamazaki, Weinstein, & Chang, 2000). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 9 Particularly, in Taiwan, Beckett and colleagues (2002) have also found that parents who have lower education, widowed, and have fewer living children are more in need of support than others in their cohort. However, empirical evidence for the relationship between the parent’s health status and receiving financial support from their children is mixed. Lee and Xiao (1998) have found that parents with difficulty in daily living are more likely than parents without difficulty to receive financial support. While other scholars have found no relationship between parents’ health and the likelihood of receiving monetary transfers from children (Hermalin, Ofstedal, & Lee, 1992; Lin et al, 2002; Yang, 1996). Likewise, older parents who are more motivated to give to those children who are more in need relative to their other children, such as those with poorer health and lower socioeconomic status (Biddlecom, Chayovan, & Ofstedal, 2002). In this type of model, older parents are more likely to receive than give to the adult children over various types of economic, emotional, and instrumental support (Lee & Xiao, 1998). Under this model, the Confucian ideals are not as significant as in the power and bargaining model, which is discussed in the next section, because the motivation to receive and give in the altruism model is based on need, rather than for the sake of keeping with tradition. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 50 Power and bargaining model. Although intergenerational transfers can be a response to actual or perceived need, there will be cases where the elderly are motivated by factors other than need, such as individual preferences. Contrasting the altruism model is the power and bargaining model, which constitutes older parents who have the economic means to live independently, but prefer to receive support from their adult children. These older parents are losing their power to extract obedience and resources from their children because they are faced with declining control over the family economy, such as their farms or businesses, as well as having skills that are quickly becoming obsolete with rapid economic development, (Lee, Parish, & Willis, 1994). Children are also absorbing new values of individualism through education and mass media especially when they move away from their rural homes to pursue higher education in urban areas or overseas in more developed countries (Goode, 1963; Nugent, 1985;Thomton, et al., 1994). Thus, parents begin to negotiate or bargain their power to receive support by making promises of future bequests and inheritance to their children. The theory of power-dependence relations proposes that an older parent's dependence is a source of power for their children (Emerson, 1972a, 1972b). A shift between power and dependence occurs between these two generations depending upon who has more control over resources. Mohn, Takahashi, and Peterson (2000) suggests that those with more power have large or rich altematives, R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 51 such as older parents who have a large kin network with more children who may be available to offer support and those with higher income, property, and wealth to offer inheritance or hire help. Further, the power network, as introduced by Cook and Emerson (1978) can also be applied in this case as widows or widowers will have more decision making power in the family and support than those who are still married because the power is concentrated upon one individual rather than two. In this model, children who seek financial help are forced to negotiate with the living parent who possesses the power, allowing the older single parents to be in a high- power network. While in a low power network, two older living parents provide more opportunity for children to negotiate with either one or the other to meet their goals. Power-and-bargaining is likely to occur more often in coresidence because there are more opportunities for bargaining and transfers to occur than generations that do not coreside together. As a result of modemization, one demographic shift that consequently occurs is a drop in fertility as parents find it more profitable to have fewer children to support. Although this is evident in Taiwan, the change in the direction of the transfers is not as dramatic as fertility decline (Fricke et al, 1994; Hermalin et al, 2002). Most parents continue to live with adult sons and receive financially from the younger generations (Liu, 1982; Lo, 1987). In the power and bargaining model, children may also use financial resources to exempt themselves from coresiding with their parents (Hermalin et al., 1990). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 52 Mutual aid model. This model emphasizes reciprocity, suggesting that the extent to which adult children help their parents is based on the parents’ past investment in them (Gouldner, 1960; Thibaut & Kelley, 1959). According to Lin and colleagues (2003), children who received a greater investment, such as tuition for education or financial gifts from parents, are more likely to provide parental support. This model is found in both the United States and Taiwan, even though support from children is given in different ways. In the US, a study by Cox and Rank (1992) shows that children who received money from their parents are more likely to visit, call, and help their parents than children who do not receive any monetary assistance. In Taiwan, children who received tuition for secondary education and above or who have received property from their parents are likely to provide financial and material support to their parents, as opposed to their counterparts who received tuition with only primary education or no property (Hermalin, Ofstedal, & Lee, 1992; Lee, Parish, & Willis, 1994). The reciprocal exchanges may involve the older parent offering help to their working adult children with baby sitting, housekeeping chores, giving of material goods (e.g., housing, food, etc.), or expenses. One of the most common forms of support provided by older persons is giving time, especially in caring for their grandchildren who have working parents. In Taiwan, 23 percent of older adults say they care for their grandchildren (Biddlecom, Chayovan, & Ofstedal, 2002). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 53 Among the older adults who provide support to their children, one-third expressed that they have encountered difficulties or worries in their efforts to provide assistance, but did not feel bothered or burdened with providing support. In return, these older parents will receive family provided services, rather than those provided by the market (e.g., hired help). The mutual aid model assumes that transfers between parents and children are due to the response of both generations transferring to each other based upon the services received by one another (Lee, Parish, & Willis, 1994). Thus, there is low adherence to normative values of support, as these older individuals who follow this model are more likely to give as well as receive from the younger generations. Corporate model. In this model, there is a strong emphasis upon the overall welfare of the family members and to maximize the their well-being and preservation through intergenerational coresidence. Members give and receive support based on the high installments of filial obligations, educational, and economic opportunities that the parent has given to their children in the past. In many recent studies on intergenerational living arrangements and support exchange pattems in Asia show that with industrialization there is a persistent receiving from their children due to the parent’s investment in their children at younger ages, though with some modifications (Casterline et al., 1991; DaVanzo & Chan, 1994; and Yang, 1992). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 54 Before industrialization and in current rural areas, the family-based economy consisted of parents involved in the children’s education or training, marriage, and employment, in order to ensure the children’s long-term commitment to the corporate unit (Ikels, 1993; Riley, 1994; Whyte, 1993). The oldest male traditionally coordinated family activities with the understanding that production benefited the group as a whole (Fricke et al, 1994). Thus, the family profits above what was paid out for taxes or rent went to support its members; while surpluses sold on the market went into a common fund and corporate decisions were made about where to invest these proceeds necessarily affected the family. Even today, Thornton and colleagues (1994) have noted the success of a family-based company or corporation that appear to be more profitable and thrive for longer periods of time when compared to new corporations that enter in Taiwan. However, with industrialization, the family economy is becoming more rare and parents find that in order to maximize the family wealth, investing in their children’s higher education will bring higher retums in the future (Lee & Xiao, 1998). These children who will then attain higher education and likely a higher socioeconomic status, are more likely to take care of their parents when they are older, if necessary (Lillard & Willis, 1997). Self-interest model. This model assumes that children provide more for parents who have the potential to give to them in the future, based upon the family wealth controlled R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 55 by the parent (Lee & Xiao, 1998). From the older person’s perspective, especially when they strongly believe Confucian ideals of filial piety, will have self-interest in giving more to the children who are wealthier and promising more to them in the future. Although these older parents are able to hire help or live independently during later years, they will live with their children out of preference, rather than need. Few studies, in the other Asian countries and the United States, however, find self-interest to be the motive for adult children’s upward transfers to their elderly parents (Lee, Parish, & Willis, 1994; Lillard & Willis, 1997; McGarry & Schoeni, 1995). During recent times, the wealth of elderly parents is generally limited in Asia. Even when adult children do have self-interest motives, their provision for wealthier parents may be more in the form of companionship and instrumental help rather than monetary goods (Lee, Parish, & Willis, 1994). Statement of the Problem Through policies that reinforce the norms of filial piety in Asia, many govemments rely heavily upon the family to take care of their older parents. However, as discussed above, the demographic and economic transitions that Taiwan has undergone during the last 50 years are associated with the country’s experience in advancements of population control measures and modemization. The anticipated consequences of these rapid transitions are forcing policymakers to face the decision of whether or not to intervene and through which mechanisms. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 56 The issue is whether or not current perceptions of filial norms held by older Taiwanese people differ from the support they actually receive from their adult children. The main question that policymakers and researchers are asking is: Are filial norms adhered to across generations a valid assumption held by the Taiwanese govemment when demographic and technological transitions have occurred? The sharp increase in the number of Taiwanese elderly is concurrent with the decline in fertility and mortality, thus placing pressure on smaller size families to provide care for the elderly. Meanwhile, there is a lack of sufficient public programs that can act as a safety net for those without traditional care providers because govemments still rely heavily on the traditional norms of filial piety with the expectation that the elderly are cared for by their children. As stated in the introduction of this proposal, the purpose of this dissertation is two-fold with the primary purpose of this dissertation is to examine how elderly perceptions of filial piety affect intergenerational exchange between older adults and their adult children in Taiwan; and the second is to use a cutting-edges statistical method of latent class cluster analysis to explore the latent structures of intergenerational support pattems that would otherwise be difficult to detect using traditional statistical techniques. In addition to the five models of motivations, the main model of motivation that is tested in this dissertation is the filial norms model. As seen in the literature, a strong sense of filial obligation is likely shared by both the older parents and their children; as parents have socialized their children into these values and R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 57 invested into their education, causing children to provide more upward transfers with a strong emphasis upon emotional support (Comman et al., 1999). Under the filial norms model, there is a hierarchical order to providing care. In Confucian societies, the oldest male son and his wife provide support to their parents. If the oldest son is not available because he is studying abroad or has deceased, then it will the second son’s duty to care for the older parents (Lin et al., 2003). If no sons are available, then it will be the oldest unmarried daughter’s duty. As for daughters, they are traditionally unreliable as caregivers for their own parents because of the social expectation for married females to leave their family of origin to become part of their husband’s family (Lillard & Willis, 1997). This principle of mandatory reciprocity of favors, renquing, implies that there is both a normative standard for regulating social exchange and a social mechanism that an individual can use to strive for desirable resources within a stable and structured social fabric (Hwang, 1987). In places like Taiwan, where children learn that the greatest shame of all is to abandon one’s parents is reinforced through the family, school textbooks, and by society at large, there is a strong filial normative obligation to provide for the parent (Lin & Fu, 1990; Wu, 1985). The second purpose of this proposed dissertation is to use latent class cluster analysis to determine the underlying pattems of intergenerational transfers that simultaneously occur between older and younger generations. Most research on intergenerational social support in Taiwan have examined either at the support received or the support given by the parent or the child separately, rather than R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 58 Biddlecom et al., 2001; Lee, Parish, & Willis, 1994; Lin et al., 2002). Further, much of the literature related to intergenerational social support in Taiwan, particularly studies using the same data set as in this proposal, has used multinomial logistic regression models, while a reliable method employed to predict outcomes for categorical dependent variables (Le, 1998), the method may not be valid for determining all possible outcomes simultaneously between the givers and receivers. The use of latent class cluster analysis can help describe the classes of intergenerational support exchange, which is potentially a methodological contribution that can be made to make predictions as to the likelihood of where individuals will fit into a typology that are otherwise undetected by traditional model, such as regression, discriminant, or log-linear analysis. Conceptual Framework As explained in the previous section, the motivations for older parents and their adult children to engage in emotional, instrumental, and economic exchanges are based upon many factors, from socio-economic status to normative beliefs. As a result of these motivations, the literature supports four possible directions and types of transfers that can occur: upward, downward, reciprocal, and none (Agree et al, 1998, 2001; Biddlecom et al, 2001; Lee, Parish, & Willis, 1994; Lin et al, 2002). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 59 Figure 1. Four Directions and Types of Intergenerational Transfers Between Two Generations. Poles of Modernization Generation: 2"°' Generation: Un-modernized Traditional/Survival No Formal Programs Types of Transfers Upward Reciprocal Exchange Fully Modernized Secular/Self-Expression Formal Programs Downward None Older Parent Economic Economic Instrumental Instrumental Emotional Emotional Economic Economic Instrumental Instrumental Types of Transfers Adult Children Figure 1 depicts the four pattems of transfers between two generations with the directions shown by the arrows and the types of transfers italicized. The types of transfers can be economic, instmmental, and emotional. The upward transfers are when older parents receive from their children, but do not provide to them. Older parents that are in this category of relationship, tend to R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 6 0 Table 1. The Intergenerational Social Support Models for Older Taiwanese Parents Based upon Economic, Instrumental, and Emotional Support. Upward Downward Reciprocal No Transfers Transfers Exchange Exchanges Description Older Older parents Older Older parents provide to their parents parents receive from children but do provide to neither their not receive their receive from children, but from them. children and their do not Parents tend to also receive children, nor provide to provide from them. provide to them. Parents tend to receive economic transfers. instrumental (i.e., babysitting) and economic help. Exchanges will be non equivalent with parents providing instrumental (i.e., babysitting services), and receiving economic transfers. them. receive economic transfers. On the contrary, downward transfers refer to older parents who provide to their children, but do not receive from them. These are older parents who tend to provide instrumental help via babysitting and economic help as well. The reciprocal exchange type is made up of older parents who both provide to and receive from their children. The type and quantity of exchanges can be non equivalent. For example, older parents may offer instrumental help to their working adult children with baby sitting, housekeeping chores, giving of material goods (e.g.. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 61 housing, food, etc.), or expenses. In return, these parents will receive economic and emotional support from their adult children. As for no exchanges, older parents neither receive from their children, nor provide to them. Table 1 provides a summary of the four directions and types of transfers mentioned above. What predicts the various pattems of intergenerational transfers? There are various possible models of motivations, as seen in the previous section regarding the literature review. However, many of the characteristics of the models overlap and are not mutually exclusive. For example, both the corporate and power and bargaining models allude to power possessed by the older parent to elicit either immediate or future support. Due to the complexity of this analysis, a cross- sectional dataset is used to examine these models motivations as groups or domains to streamline the various models with overlapping characteristics. As explained in the last chapter of this dissertation, using cross-sectional data inevitably limits the models available for testing future support that can be received by the older adults, such as in self-interest and corporate models. Table 2 shows the models of motivations encased in the heavy black lines, which predict intergenerational transfers (indicated on the left hand column). The one model that is left out of the box is the estrangement model, which can exist though it is the antithesis of any motivation because older adults and their children have no relationship besides biological ties. In this model, neither parents nor their children gives nor receives from each other because, besides biological attachments, older parents have no geographic, emotional, economic, or social attachments to their adult children. For R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 62 example, working adult children may live abroad and do not contact their healthy older parents who prefer to live economically independent and disregard traditional norms of filial piety. The most basic domain of motivation for adult children to provide for their older parent is based upon physical or economic needs. Adult children, who are motivated by the economic and health needs of the parent, may provide upward transfers to their older parents, which follows the altruism model. The types of upward transfers can be economic, emotional, or instrumental, depending on the needs of the parent. Hence the first domain introduced is need, which motivates adult children to provide for their older parent who have poor health or low-income levels. The second domain of motivation is based upon the exchange that occurs between older parents and their children. Just as meeting basic needs for survival can be a natural and instinctive behavior of adult children, society teaches that economic and social exchanges take place as people reciprocate favors, gifts, products, or services to each other, in order to meet their own needs. Thus, giving and receiving various forms of support may become a visible relationship, especially in newly modernized societies, such as Taiwan. Under the mutual aid model, reciprocal exchanges can take place because adult children give upwardly to their parents through various types of care, in return for the older parent’s help, such as caring for their grandchildren. Often times, this occurs among adult children, especially working daughters, who have attained higher educational levels due to R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 63 their parental investment in the past to achieve a higher socio-economic status (Lee & Xiao, 1998). However, current reciprocity does not necessarily have to be based on a history of investments made by either generation. An older parent can give immediate help to working adult children who need help with babysitting their children or financial support. On the other hand, there are older adults who have made investments in the past to support the child through buying a house, paying for tuition, and paying medical bills. Hence, the exchange domain is examined using both these variables operationalizing past and present investments of support. Meeting intergenerational needs through social support exchanges is the next logical step to include in the motivations for social support models. The next domain is based on the resources of power that an older adult possesses to intentionally or unintentionally elicit to type of social support behavior. This is a step beyond meeting needs of either generations, but more about planning for possible future needs. There are three models of motivation with two types of transfer pattems that are predicted under this domain. There can be the promise of future inheritance of property and assets from older parents that motivate their adult children to provide economically, emotionally, and instmmentally for them, as is the case of the power and bargaining model. The self-interest model is renamed to income generator, which portrays an older adult who is the main wage eamer and having dependent adult children who have higher education and potentially more assets to provide more for them in the future. Although these older adults are possibly providing to their adult children for insurance purposes—tapping R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 64 into their children as support resources when they need it most in the future—the underlying intentions and the future results of these actions are unknown because self-interest is not directly measured. Thus, in this model, the net transfer is downward flow. Another type of model that has downward flows is the corporate model. In this model, the family is like a business corporation, with the head being the oldest member who makes decision to maximize the joint wealth of the entire family, so that every member benefits from the corporation. Thus, the older parents will redistribute wealth to the younger members whom they expect greater returns in the future. The ideology of the parent is the next domain of motivations because the older adult is expecting the manifestation of societal normative values in the social support received from their adult children. As predicted by the filial norms model, those who have high levels of filial expectation will have children who produce the outcome of upward transfers as well. In this model, older parents who adhere strongly to filial piety will be less likely to give to their adult children as they expect to receive from them. This is the final model that is introduced and is also the main model of interest as filial norms include the needs, exchanges, and power of the older adult. Under the lens of the modified modernization framework, there is a modernization continuum based upon changes in cultural attitudes, development of formal programs, and intergenerational transfer pattems. In Figure 1, the two extreme poles of modemization are shown. The unmodemized countries display R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 65 mainly traditional normative values that geared more towards caring about one’s livelihood and survival. Hence, the pattern of transfers is mainly upward from the younger generation to the older generation to ensure the survival of the elderly in the absence of formal programs. In Asia, examples of unmodemized countries are Cambodia and Laos. The other extreme are fully modernized countries that have prevailing societal attitudes of self-expression and secular-rational orientations. These fully modernized countries have a long history of developed formal programs to support the elderly, with more downward flows of support that older parents give to support their children. This is found in many European countries, such as Sweden and Europe. Taiwan can be seen as a partially or newly modemized country that sits in between these modemization poles, with values in between secular-rational and self-expressive. While intergenerational transfer patterns are moving from upward transfers (adult children giving to older parents) to reciprocal exchange (adult children and older parents giving to and receiving from each other). Although Taiwan is a newly modemized country, the pattems of transfer existent in families is not yet in the downward direction as in the United States and Europe. Hence, the model of motivation that is likely to be existent in Taiwan is mutual aid as parents have made past and present investments in their children, such as giving them R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 6 6 tuition assistance for education and providing support by offering babysitting services, they receive economic and emotional support in return. Table 2. Description of Models of Motivation for Intergenerational Transfers within the Matrix of Domains of Motivations by Directions of Transfers. _____________________ Domains of Motivation Directions of Transfers Need Exchange Resources of Power Ideology Upward Altruism Older parents receive transfers from adult children based upon their own economic and health needs. Power and Filial Baraainina Norms Older parents Older receive transfers parents from adult children receive based upon the transfers promise of future from inheritance of children property and based assets. upon the strong adherence to filial piety. Downward Income Generator Older parents who are the main wage earners in the household give to their adult children who are highly educated and wealthier, expecting future returns. Corporate Control Older parents have the economic decision making ability in the household to give to the adult children whom they can expect the greatest amount of future returns. Reciprocal Exchange Mutual Aid Older parents receive transfers from children in whom they have invested more in the past and present._______ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 67 Research Questions 1. What are the major pattems of exchanges that take place between an older parent and any of their adult children? 2. What are the motivations for older adults to engage in social support exchanges with their adult children? a. Do levels of filial expectations matter in the types of exchange relationships that an older adult is engaged in? b. What kind of exchange relationships does an older adult engage in based upon their economic and health needs? c. Is there a positive relationship between engaging in exchanges and the investments made in children through buying a house, paying for tuition, and paying for medical bills? Likewise, does the same relationship exist when older parents provide childcare and monetary support to their adult children? d. What kind of social exchange relationships do older parents engage in who have the following three types of economic power: promising future bequests of property, corporate control, and income provision? 3. What are the characteristics of older parents and their adult children who are unlikely to engage in any exchange of support? R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 6 8 Hypotheses Taiwanese families may have diverged from the traditional models of adult children giving instrumental, economic, and emotional care to their older parent because of modemization. Therefore, there may be latent forms of intergenerational transfers that are difficult to observe, both theoretically through the literature and methodologically through traditional techniques of cluster analysis, which are proposed in the sub-section on intergenerational transfers pattems. The main hypothesis is that the older parents perceive filial norms as the main motivator for transfers in Taiwan, but due to the effects of modemization, traditional pattems of transfers are now modified to be mainly economic types of transfers. Competing hypotheses for models of motivations, based upon the types and directions of transfers, are also proposed. Hypothesis 1. There are four general intergenerational transfer patterns that occur between the older parents and their adult children: reciprocal transfers, upward transfers, downward transfers, and no transfers. Summaries of these types of transfers are found in Table 1 . Reciprocal transfers occur when older parents provide to their children and also receive from them. These transfers may or may not be equivalent or balanced, as R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 6 9 economic transfers. Downward transfers occur when older parents provide to their children but do not receive from them. In these types of transfers, parents tend to provide instrumental (i.e., babysitting) and economic help to their children. Upward transfers occur when older parents receive from their children, but do not provide to them. Finally, there are the no exchangers who are older parents neither receiving from nor providing to their children. Hypothesis 2. There are four domains of motivation for intergenerational transfers proposed for Taiwan: need, exchange, resources ofpower, and ideology. Linking the modemization level and reciprocal exchange pattems to the model of motivation, it is expected that Taiwan will have a predominance of mutual aid motivations for older adults to continue their investments in their adult children, in retum for their children’s support. There are three models of motivations related that can be inferred among dyads engaging in upward transfers: altraism, power and bargaining, and filial norms. In the altmism model, older parents mainly receive transfers from their adult children based upon their own economic and health needs. The type of upward transfer will depend upon the need of the older parent. Parents who are older, female, single, in poor health, and have no pension are likely receiving economic, instramental, and emotional support from their adult children. In the power and bargaining model, older parents will receive transfers from their children with the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 7 0 promise of future bequests of property and assets and with the implied threat of the fact that wealth will not be transferred to non-compliant children. Those with highly educated children will tend to give in upward transfers to their parents with the potential of receiving a higher portion of inheritance from their parents in the future. Thus, parents who have either partially or not yet divided property, are likely receivers of support from older sons who are highly educated, working, and have at least one grandchild. In the filial norms model, older parents who adhere more strongly to filial piety norms will more likely receive transfers from their children, but in a modified form where upward transfers of financial support dominate over other types of support, with low amounts of downward transfers. The motivations behind downward transfers are of two types: income generator and corporate control. The first downward transfer type is the income generator model, which consists of older parents who have independent income and provide for the household as incentive for highly educated adult children to provide transfers to them in the future. However, older parents with younger children who are pursuing higher levels of education will mainly be giving downwardly to them as an investment to receive future support. Older married adults who are higher socioeconomic status with altemative economic resources other than the adult children and in good health are likely giving to a coresident older married son, who is highly educated and working full time. The corporate model consist of older parents who are the main economic decision makers in the household, even though they may not be a primary income eamer for the family, who give to younger adult R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 71 children with lower education and unemployed because of financial needs. Thus, older adults who have children in need with lower socioeconomic status, younger, and single are more likely to receive downward transfers from their older parents. The only model of motivation that predicts reciprocal exchanges is called mutual aid, which is divided into immediate and past investments. In the former type, the older parent is providing support to children in need of these services or monetary supplements. The latter consists of older parents who will give and receive transfers from children in whom they have invested more through paying for education, housing, and medical bills. These older parents are likely to still be in good health and engage in reciprocal exchanges by offering babysitting services and money to adult children who are highly educated, working daughters, and have children. The estrangement model predicts no transfer pattems will occur between generations. Besides biological attachments, older parents neither give to nor receive fi-om their children due to the lack of economic, geographic, social, or emotional attachments to each other. There are two profiles of these older parents who fit into this model. The first are parents who likely to be younger, healthier, having a higher socioeconomic background, having low adherence to filial norms, and have highly educated, noncoresident children who are likely to live abroad. The second are parents who are older, in poor health and have no contact with their children are likely to be older, either having a high socioeconomic status because R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 72 they have property or assets to afford services, or a lower socioeconomic status where they are receiving a government subsidy. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 73 CHAPTER FOUR Research Design and Methods The purpose of this chapter is to outline the research design for this dissertation. First, the study sample will be described in detail. Next, a description of the instruments selected to measure the dependent and independent variables will be outhned. Finally, a description of the following two statistical methods chosen for this analysis: 1) latent class cluster analysis and 2) multinomial logistic regression analysis. Data Source In order to study the age 65 and older population in Taiwan, this dissertation will use data from the fourth wave of the Survey of Health and Living Status of the Middle Aged and Elderly in Taiwan (N=4,440), which was collected in 1999. This is a nationally representative dataset collected by the Bureau of Health Promotion and Protection (BHP), formerly known as the Taiwan Provincial Institute of Family Planning (TPIFP), under the direction of Ming-Cheng Chang, as a joint survey project with the Population Studies Center at the University of Michigan with Albert Hermalin as the principal investigator (Hermalin et al, 1989). Funding for the survey project has been provided by both Taiwan and the United States. Among the Taiwanese government agencies that have provided grants for this project, are the National Health Department, the National Health Research Institute (Grant No. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 74 DDO1-861X-GR60IS), and the former Taiwan Provincial Government. The United States National Institute on Aging (Grant No. R37 AG07637) also provided supplemental funding as part of four-nation comparative survey project entitled “Rapid Demographic Change and the Welfare of the Elderly,” based at the University of Michigan and conducted in collaboration with population institutions in the Philippines, Singapore, Taiwan, and Thailand. The sample at the first time point of data collection consisted of 4,049 respondents age 60 and over in 1989, reflecting the socio-economic, ethnic, and gender distribution of older Taiwanese. In 1993, the next wave of the survey was carried out with 3,155 of the living respondents from baseline. In addition, one- quarter of the sample were randomly selected to have their living children participate either in face-to-face interviews for coresident children with their spouses or telephone interviews were attempted for noncoresident children with their spouses (Roan, Hermalin, & Ofstedal, 1996). There were 576 coresident children and their spouses and 1907 noncoresident children with their spouses who were interviewed, which is a total of 2483 respondents who were added to the study. In 1996, two questionnaires were administered to replenish the longitudinal sample: one for those age 50 to 66, and one for those age 67 and above. The interviews were conducted in- person between April and November of 1996. The 1996 sample size for the original panel from 1989 was 2,669 respondents; the second panel (ages 50-66) had 2,462 respondents. The two panels were combined and properly weighted to create a representative sample of the Taiwan population ages 50 and over. The 1999 total R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 75 survey sample included 4,400 older respondents with 2,310 original older respondents from baseline. For the type of statistical analysis selected, only one wave of the data collection is used to estimate the static latent variables. Thus, in order to make generalizable conclusions about the current generation of elderly in Taiwan, the most current data available is used, which is from 1999. The sub-sample of 2,763 older people who are sixty-five and older are used in this analysis. Each respondent is matched to their responses about each of their coresident and noncoresident biological children, which produces a total of 9,477 dyads. On average, each respondent had five living biological children, with a minimum of one and a maximum of 12 (Table 6). Measures This section discusses the choice of variables used to operationalize the concepts of the hypotheses introduced in chapter three, the questions asked in the surveys regarding theses variables, and their levels of measurements. Tables 3 and 4 lists each variable, the question, and the response categories, but are separated into two tables that present the independent and dependent variables. Tables 5 and 6 consist of frequencies, which include percentages, means, and standard deviations for the variables used in this dissertation and other background characteristics related to the respondents. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 76 Table 3. Table o f Independent Variables Used in the Analysis Characteristics Question Variable Socio-demographic Characteristics of Older Taiwanese Respondent Age Derived from Birth year Recoded into two dummy variables: l=Age 65-74 l=Age 75-84 Reference=Age 85+ Gender Respondent's reported gender l=Female 0=Male Ethnicity Are you Fukienese, Hakka, Mainlander, Indigenous, or where are you from? Recoded into l=Fukienese l=Hakka l=Mainlander Reference=Indigenous Economic Need: Economic Resource Literate What is your highest education level? Response categories 00=llliterate, 1- 17+=Elementary to Post-College levels, and 90=Literate Recoded as: l=Yes(l-17+, 90) 0=No (00) Social Need: Marital Status Is respondent currently married or living with a partner or spouse? l=Yes (married or living with a partner or spouse) 0=No (single, divorced, or separated) RuralAJrban Residence Is respondent currently residing in a rural area? l=Rural (includes farm) 0=Urban (Large/small city. Town) R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 77 Table 3. Table of Independent Variables Used in the Analysis (continued) Exchange: Support Provisions Economic Financial Have you given children financial help? l=Yes 0=No Instrumental Babysitting Do you babysit any of your grandchildren? l=Yes 0=No Need Health Need: Self- reported Health Regarding your state of health, do you feel it's: l=Poor 2= Not so good 3=Average 4= Good 5= Excellent Recoded as: l=Good to Excellent Health 1= Average Health Reference=Poor to Not so good Health Economic Need: Social welfare program In the past 12 months, did you receive any income from a social welfare program? l=Yes 0=No Corporate Group Model: Parent is Family Head Meeting Needs of Children Economic Decisions Respondent and/or spouse (as opposed to children) in this household usually makes the final decisions on major economic issues such as buying or selling something of large value or making a major purchase. l=Yes 0=No R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 78 Table 3. Table of Independent Variables Used in the Analysis (continued) Resource of Power Division of Property Have you and your spouse distributed all of your properties among your children? Have you and your spouse distributed part of your properties among your children? Have you and your spouse have no property to divide? l=Yes, all divided 1= Yes, partially divided 1= Yes, no property to divide Reference=Not yet divided Economic Decisions Respondent and/or spouse (as opposed to children) in this household usually makes the final decisions on major economic issues such as buying or selling something of large value or making a major purchase. l=Yes 0=No Past Investments Made in Children Types of money transfers that have been made to children In the past three years, have you/your spouse given a large amount of money or property to your children? For what? Buy a house? l=Yes 0=No Education abroad? l=Yes 0=No Medical treatment? l=Yes 0=No Parent’s Own Resources Provided to Children Main Wage Eamer for Household Respondent is the main breadwinner for this household l=Yes 0=No R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 79 Table 3. Table of Independent Variables Used in the Analysis (continued) Filial Norms Model: Social Power Filial Expectations Scale Taken as a composite score of the following 10 items listed below: Total score ranges from 1 (low filial expectations)-50 (high filial expectations) Once children have started families, parents should make every effort not to live with them. Elderly people should be economically independent, rather than dependent on their children When children need it, parents should help them take care of their children (the grandchildren). The government already takes good care of the elderly If an old woman’s husband has been dead for a while, would you approve of her remarriage? 1= Very much agree 2= Agree 3=Neutral 4= Disagree 5= Very much disagree R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 8 0 Table 3. Table of Independent Variables Used in the Analysis (continued) Filial Expectations Scale (continued) Items that have been reverse coded. When parents get old children should live with them. Young people today do not respect elderly people as much as they did in the past. Old people should keep some savings or property for themselves or the children will not respect them as much. If an old man’s wife has been dead for a while, would you approve of his remarriage? 1= Very much disagree 2=Disagree 3=Neutral 4=Agree 5= Very much agree Sociodemographic Characteristics of Children Age of child Age of adult child Recoded into three dummy variables 1=19 through 34 1=35 through 54 1=55 and over Reference=12 through 18_____ Birth order of child Adult child is only child, first bom, etc. Recoded into three dummy variables: l=01dest 1= Youngest child I=In-between Reference=Only child R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 81 Table 3. Table of Independent Variables Used in the Ana ysis (continued) Gender Respondent's reported gender l=Female O^Male Place of Residence Distance of child to parent Recoded into 4 dummy variables: l=Lives next door l=Lives same neighborhood l=Lives elsewhere in Taiwan l=Lives overseas Reference=Coreside Years of education For those who are 6 years or older, how many years of education have you completed? Recoded into two dummy variables: l=High school graduate 1= Some college or more Reference= Less than high school Employment For those who are 6 years or older, what is the child’s employment status? Recoded into two dummy variables: 1= Working full time 1= Working part time Reference=Not working Marital status For those 15 years or older, what is the child’s marital status l=Married or living with a partner 0=Single Grandchildren For those 15 years or older, how many children does adult child have? 1=1 or more 0=None R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 82 Table 3. Table of Inde pendent Variables Used in the Analysis (continued) Frequency of face- to-face contact How often do you see him/her? Recoded into four dummy variables: l=See everyday l=See every few days to weekly l=See every few months to monthly l=See every few years or yearly Reference=Live with Frequency of phone contact How often do you talk on the phone with him/her? Recoded into four dummy variables: l=Talk everyday l=Talk every few days to weekly l=Talk every few months to monthly l=Talk every few years or yearly Reference=Live with Dependent Variables The social support measures are the indicators for the latent class cluster analysis, which are presented in Table 4. The clusters that form from the analysis are then used as the dependent variables in the multinomial logistic regression. In this dataset, there are two types of measures for support: 1) giving support, which is from the perspective of the older parent who gives to the adult children; and 2) receiving support that is also based upon the older parent’s response R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 83 of which adult child they receive various types of support from their parents. The frequency distribution and percentages of older respondents who give and/or receive the support items asked in this survey are listed in Table 5. Table 4. Table of Dependent Variables Used in the Analysis Support received by Older Parents from Adult Children Emotional Do any of your children listen to you talk about your worries and concerns? l=Yes 0=No Economic Material Financial Do any of your children send you personal belongings, such as food, clothes, or other material supports to accommodate the needs of your daily life? Do any of your children give you financial help? l=Yes 0=No Instrumental Do any of your children give you help with light household chores like cooking, laundry, sweeping the floor, cleaning, dish washing, and grocery shopping, in your household? l=Yes O ^ ^ N o Support received by the older parent from the adult children. The measures operationalizing receiving support encompass emotional, economic, and instrumental types of assistance. The emotional assistance received by the older parent from the adult children includes the following two questions: “How much do you feel that you family, relatives or friends are willing to listen when you need to talk about your worries or problems? Would you say a great ^. In a comparison of Taiwanese parents’ and children’s reports o f the transfers children made to their parents, using the 1996 wave o f the Survey o f Health and Living Status o f the Middle Aged and Elderly in Taiwan. Roan, Hermalin, and Ofstedal (1996) found a relatively high degree of concordance in reports (upwards o f 65%) for the following three types o f support: money, help with R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 84 deal (=1), quite a bit (=2), some (=3), very little (=4), or not at all (=5)?” Respondents were allowed to indicate up to two possible children that they thought were most willing to listen, which were coded by birth order and gender as in the measures operationalizing support given to adult children by the older parent. The responses were recoded as one through four for those who are in the group of interest that have children who listen to the older respondent’s worries or problems (“yes”=l). The same coding was used for another question related to emotional assistance received from the adult children: “How much do your family, relatives or friend make you feel loved and cared for?” The response categories also ranged from a “great deal” (=1) to “not at all” (=5), with those who answered the last category as the reference group. Table 5. Frequency Distribution and Percentages of Support Indicators Indicators n % Support Given from Parent to Adult Child^ Give Babysitting 939 9.9 Give Money 221 2.3 Give Daily Assistance 16 0.2 Give Material Belongings 20 0.2 Support Received from Adult Child to Parent Receive Love and Care 3489 36.8 Receive Listening 2602 27.5 household chores, and visits. These same measures are used in this dissertation from the 1999 wave, but only from the parent’s perspective. ^ There are two other questions asked o f the respondent concerning the support they have provided to their adult children, which were not included in this analysis because the distribution was too sparse With only 0.2% o f the total sample that responded “yes” to the following two questions: “Do you help someone in your family with daily life activities? (Those who answered “Yes” were then asked which child they provided this assistance to); and “Do you regularly send belongings like food or clothes to anyone?” (Those who answered “Yes” also reported which child they provided this assistance to). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 85 Table 5. Frequency Distribution and Percentages of Support Indicators (N=9,477) (continued) Receive Daily Assistance 185 2.0 Receive Material Belongings 929 9.8 Receive Money 1851 19.5 Receive Care when 1 1 1 3489 36.8 Similar to variables given from the older parent to the adult children, the economic support variables consist of financial and material assistance. Respondents were asked: “In the past year, has your child or relative given you or your spouse money for your daily living expenses?” Respondents were then asked to name up to eight possible adult children who provided financial assistance. As with the same coding as the variables above, the responses were coded by birth order and gender of the child. For material belongings, the following questions was asked: “Does anyone regularly send you personal belongings such as food, clothes or other material support to accommodate the needs of your life (presents for holidays, birthdays or socialization are not included)?” Respondents were asked to name up to four possible people, which were coded by their relationship to the respondent, with only the adult children codes selected for this analysis. The instrumental support received from the adult children is in the form of light household chores. The question was asked: “Who takes care of most of the “light” household chores like cooking, laundry, sweeping the floor, cleaning, dish washing and grocery shopping, in your household?” Respondents then specified the relationship of the person who performs this assistance and those who indicated adult R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 8 6 children were recoded as “yes,” which was the group of interest. However, due to the sparseness for the distribution of this variable across the sample, the older parents who received daily assistance were dropped from the analysis. Table 6. Distribution and Means of Respondent Sociodemographic Characteristics in 1999 ____________________________ _______________ Characteristics Percentage/Mean S.D. Sociodemographic Background Age (Min=65, Max=98) 74.5 6 Gender Male 53.2 Female 46.8 Ethnicity Holo 62.9 Hakka 15.6 Mainlander 19.9 Indigenous 1.3 Place of Residence Rural (includes farm) 33.7 Urban (includes cities and towns) 66.3 Marital Status of Respondent Married or living with a partner or 60.6 Spouse Single, divorced, or separated 39.4 Live with spouse Yes 62.1 No 37.9 Self-reported health Excellent 9.1 Good 19.5 Average 34.0 Not so good 29.6 Poor 8.6 Economic Resources Literacy Literate 64.7 Illiterate 35.3 Pension or Retirement Fund Yes 87.7 No 12.3 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 87 Table 6. Distribution and Means of Respondent Sociodemographic Characteristics in 1999 (continued) Division of Property N ot yet divided 53.8 All divided 11.9 Partially divided 10.4 No property to divide 24.9 Economic decisions Respondent and/or spouse makes 58.9 final decision Children makes final decision 41.1 Income Levels Lower income quartile Median income quartile Higher income quartile Main wage earner Respondent is the main wage earner Respondent is not the main wage earner Nonfamily Resources Enrolled in Social Welfare Program During past 12 months Yes 39.1 No 60.9 Sociodemographic Background o f Respondent’s Children Number of sons (Min=0, Max=8) 2.3 1.4 Number of daughters (Min^O, 2.2 1.6 Max=10) Gender Female 66.8 Male 33.2 Age Group (years) 12-18 7.9 19-34 24.5 35-54 59.8 55 and over 7.8 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 8 8 Table 6. Distribution and Means of Respondent Sociodemographic Characteristics in 1999 (continued) Birth order Oldest 37.1 In-between 45.4 Youngest 12.1 Only Child 5.4 Coresidence Lives with ehild 30.0 Does not live with child 70.0 Number of adult children coresiding 1.0 0.8 with respondent (Min=0, Max=6) Live with Unmarried Children Yes 21.8 No 78.2 Live with Married Son Yes 51.3 No 48.7 Live with Daughter-in-law Yes 47.1 No 52.9 Live with Married Daughter Yes 6.0 No 94.0 Live with Grandchildren Yes 56.5 No 43.5 Other Places of Residence Lives next door 11.7 Same neighborhood 19.7 Lives elsewhere in Taiwan 36.7 Lives overseas 1.9 Past Investments Received Either buying a hOuse, tuition for 20.8 education abroad, or medical treatment Neither buying a house, tuition for 79.2 education abroad, nor medical treatment R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 89 Table 6. Distribution and Means of Respondent Sociodemographic Characteristics in 1999 (continued) Education Highest level of education for child 10.6 3.7 living with respondent (Min=No formal schooling, Max^^Postgraduate degree) Highest level of education for child 11.8 3.8 not living with respondent (Min=No formal schooling, Max=Postgraduate degree) Work Full time 75.9 Part time 0.2 Unemployed 23.9 Marital Status Married or living with a partner or 72.8 spouse Single, divorced, or separated 27.2 Grandchild Has at least 1 89.2 None 10.8 Frequency of Face-to-face Contact Daily, coresidence 36.9 Daily, noncoresidence 10.6 Weekly, noncoresidence 21.4 Monthly, noncoresidence 27.5 Yearly, noncoresidence 3.6 Frequency of Phone Contact Daily, coresidence 36.4 Daily, noncoresidence 4.5 W eekly, noncoresidence 31.2 Monthly, noncoresidence 14.5 Yearly, noncoresidence 13.4 Filial Normative Expectations Levels of normative expectations 32.2 3.54 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 9 0 Independent Variables Motivational factors. Based on the microeconomic theories and models of social support exchange, there are many motivating factors for older parents and their adult children to give and receive from each other. Conceptually, there are four main domains that are tested, under which these motivational factors fall; need, exchange, resources of power, and ideology. The need domain includes health and economic needs of the older person under the altruism model. While the exchange domain is comprised of the current and immediate provisions and the past investments made by the older parent for their children, which are temporal components of the mutual aid model. The resources of power domain represent three components under the power and bargaining model, which is the parent’s control over their property, economic decision making, and income that they provide for the household, in order to receive support or no support from their children. The latter is possible if the older parents have these resources of power, which means they are likely economically self- sustaining. The final domain tested in this analysis is ideology, which is represented by the levels of filial expectations perceived by the older person. The models are built in a hierarchical fashion, building on similar characteristics from the previous models while adding a unique variable each time to distinguish a type of motivation. The determination of which predictors to use for each model of motivation is contingent upon the typology of transfers that exists in the population. Thus, this section will present only the independent variables for each model of motivation R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 91 based upon the existing typology found from conducting the latent class cluster analysis (see next chapter). The baseline model of motivation is the need domain, which consists of two components of the altruism model: health and economic needs. As indicated from the literature review in the previous chapter, there are two types of needs: health and economic. The variables used to operationalize these concepts are self- rated health and welfare, respectively. The self-reported health score is coded as those with not so good to poor health being the referent group (“poor’-O), and each of the other three categories are coded as separate groups of interest with average (“average”=l) and good to excellent (“excellent’-1). In terms of self-rated health, only 8.7% thought their health was excellent, 19.1% rated their health as good, 33.8% of the sample rated themselves as average, and another 29.8% felt that their health was “not so good.” The older parents with worse health will have more needs and may enlist their children to help them. The older person’s need for economic support is based upon their whether or not they are enrolled in a welfare program, as they are likely to need support from their children. The question is asked in the following way: “Have you had social welfare relief fund, government subsidy, low income subsidy, etc. as a major source of income in the past year?” (“Yes”=l and Others are recoded to=0). The next domain is exchange, which describes the immediate and past investments of mutual aid. The older parent who makes instrumental and economic provisions measures the immediate or current exchange process. Based R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 92 upon the response codes corresponding to the person receiving this support, those who report on adult children are selected for this analysis. For the instrumental giving measures, the following question asked of the older respondent is used in this analysis: “Do you or your spouse provide help with babysitting^ (Those who answered, “Yes” were then asked a two-part question: whether it was “You” or your “Spouse” who provided this service and to whom you provided this service to). There is also a measure for economic support provided to the adult children which is asked of the older respondent: “In the past year, have you or your spouse given money to your family members or friends for expenses other than educational expenses?” (Those who answered “Yes” were then asked a two-part question: whether it was “You” or your “Spouse” who provided this service and to whom your provided this service to). For each these questions, respondents who do provide the type of assistance are the group of interest (“yes”=l) and those who do not belong to the reference group (“no’-0). For the past investments made by the older parent within the mutual aid model, these include providing a house, tuition for education abroad, and medical treatment during the last three years. The following question was asked: “Have you given a large amount of money or property to parents, children, grandchildren or other relatives to help them buy a house, go abroad to study, and/or obtain medical treatment?” Each variable is dichotomized with 1= “ Yes” (group of interest) and 0= “No” (referent group). If the variables operationalizing support provided by the older parent to the adult children predict upward transfers, then the mutual aid model is inferred in the Taiwanese population. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 93 The resources of power domain are operationalized by the availability of property, control over economic decisions, and income generating abilities of an older adult. It is hypothesized that this model can predict the following two types of upward transfers, depending upon which of the three variables are significant: 1) Those who have some or all of their property and have decision making power regarding the distribution of money are likely to receive emotional, economic, or instrumental support because they are commanding immediate upward transfers with the promise of future bequests; and 2) Children who are dependent upon the older adults as the main wage earner for the household are not giving any upward transfers. The first sub-hypothesis under this domain alludes to older parents who have potential assets and property to give to their children, but require their children to take care of their needs. Measures for economic resources include property of the older person which is indicative of the power they still have in either commanding respect and support from their adult children in the present moment with the promise of bequests made for the future (Kohli, 1999; Lee, Parish, & Willis, 1994). The first measure used to operationalize this model is whether or not the respondent has potential property to give to their child, in order to solicit their help. “Respondents were asked whether they have distributed part or all of their properties among their children?” Each response category was recoded as dichotomous variables with “no property to divide” (=0) as the referent group and the following three variables of interest: “those who have not yet divided any property” (=1), those R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 94 who have “partially divided their property” (=1), and those who have “divided all their property” (=1). The next variable operationalizing the resources of power domain is inquiring whether or not the older adult is the major economic decision maker in the household. The following question is asked: ’’Who in this household usually makes the final decisions on major economic issues such as buying or selling of large value or making a major purchase?” (with “respondent”=l and “others” are recoded to=0). The second sub-hypothesis under the resources of power domain predicts downward transfers based upon the older person’s resources of power as the main wage earner. Detailed questions on the older respondent’s income resources entailed asking whether or not they have received a pension or retirement income during the last 12 months with those who answered “yes” as the group of interest (“yes”=l) and those who answered “no” as the referent group (no=0). Another question used to operationalize self-interest is: “Who is the main breadwinner for this household?” (“Respondent”=l and “Others” are recoded to=0). The last model, which is the main model of interest, is the filial norms model, which tests the levels of filial expectations. There are ten questions that are asked in the survey (see Table 3), which range from 1= “Strongly agree” to 5= “Strongly disagree.” One of the items of the scale is the expectation of the children, which is asked of the older respondent: “When parents get old, children should live with them.” Respondents were also asked about their own expectation of coresidence and financial situation by answering how much they agree with the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 95 following two questions: “Once children have started families, parents should make every effort not to live with them,” and “Elderly people should be economically independent, rather than dependent on their children.” Five of the items were reverse coded so that 1= “Very low filial expectations” to 5= “Very high filial expectations.” In particular, there are three items that may seem unrelated to filial expectations, but these are under the broader Confucian norms of filial piety (Chow, 2002). The first is if those who very much disagree with the statement that “The government already takes good care of the elderly,” then this implies that the older person has high filial expectations as the children fulfills this duty of care better than the government (Zhang, 2003). The next statements are related to remarriage by gender. “If an old woman’s husband has been dead for a while, would you approve of her remarriage?” In Confucian culture, the widow is shunned from remarrying as she has the obligation to remain loyal to her family and children must remain respectful toward their deceased father by ensuring that their mother remains single (Marsh, 2002). On the other hand, a different standard is applied to the widower in this paternalistic society, where children should respect and support their older male parent if they so choose to remarry. Hence, higher filial expectations are associated with those who very much agree with the following statement: “If an old man’s wife has been dead for a while, would you approve of her remarriage?” After scaling the ten-items, the scale ranged from 1 through 50, a Chronbach alpha score of .80 was obtained, which shows high reliability of the filial expectations scale. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 96 Sociodemographic characteristics of older respondent. Age of the older parent will be statistically controlled for in this analysis because there is some evidence of significant increases in the amount of care received by adult children when parents grow older (Lee, Parish, & Willis, 1994; Lee & Xiao, 1998; Sun, 2002). However, in longitudinal studies, age of the parent does not seem to be significant in longitudinal studies measuring the likelihood of receiving financial support, as reported from 1989 to 1999 by Lin and colleagues (2002). Since this dissertation is a cross-sectional study, age will be controlled for as indicated by its significance as a predictor in other cross-sectional studies in the literature (Zimmer et al., 2003). Age is categorized into three dummy variables whether or not the older person is “65 to 74 years old” (=1), whether or not the older person is “75 to 84 years old” (=1), and whether or not the older person is “85 years and older” (-0), which is the referent group. The mean age of the sample is 74.7 with a standard deviation of 6.03; the youngest respondent age is 65 and the oldest is 98. Gender is considered to be a potentially confounding variable in studies of social support, and therefore is statistically controlled for within this analysis. The gender variable is coded with females as the group of interest (“female’-1) and males as the referent group (“male”=0). Various studies have shown that females tend to report higher expectations of future support from children than women, although this number is declining (Chang & Ofstedal, 1991; Chen & Jones, 1989). Therefore, this study will control for gender to determine the effect of R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 97 filial expectations and motivations on social support exchange patterns. The gender distribution at baseline is comprised of 50.9% men and 49.1% women. The three ethnic groups specified in the analysis are the Holo, Hakka, and Chinese, and comprise 67.2%, 17.2%, and 13.9%, respectively. The marital status of older Taiwanese adults is also important and similar to Western literature that links the status of widowhood with receiving more from children (Weinstein et al., 1994). In this study, the marital status is a coded as married or living with a partner is the group of interest (“married’- 1) and being single, widowed, or divorced is eoded as the referent group (“unmarried”=0). Respondents who are married or living with a partner comprise 60.6% of the sample. The ethnicity variable has also been shown to be of important relevance to the studies in Taiwan. The ethnic group is coded with the indigenous minority as the referent group compared to each of the three other ethnic groups as separate dichotomous variables: “Holo” majority group (=1), “Hakka” (=1), and “Chinese” (=1). As mentioned in the literature review, the subculture of the family within each ethnic group can determine a different type of social support exchange pattern. For example, the one million older Chinese men who immigrated to Taiwan in 1949 with the KMT army who have remarried are included in this analysis and are likely to have adequate pension, which will be an interesting group to examine. Also, of interest to demographers are immigration patterns that occur in society, the inclusion of the Chinese group shows the differences, if any, between the native Taiwanese and the Chinese who have immigrated from China. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 98 The rural/urban residential difference is an important modemization factor to include in this analysis as Taiwan has rapidly changed in the last fifty years, the implications of the urban life upon the manifestation of filial piety, particularly older individuals living in the rural areas has been a well-documented problem in literature. The inclusion of this variable is to control for any rural/urban differences in determining the effects of filial expectations and motivations on intergenerational support exchange patterns. The variable of interest is rural living (“rural’- 1) and urban living is the referent group (“urban”=0) as 60% of the total population currently lives in urban areas (GIO, 2002). The socio-economic status of the older adult is operationalized by both education and income. Studies have shown that the socio-economic status of older people in Taiwan is strongly correlated with education, income, or occupation (Anderson & Armstead, 1995; Williams & Collins, 1995; Zimmer et al., 2003). However, in this analysis, the average level of education for the sample is 3.93 years, with a standard deviation of 4.35, based on a range of zero to seven years of education. Since 60% of the sample has not attained education, whether or not a person can read is used as a proxy with those who are literate as the group of interest (“literate”=l) and those who are illiterate as the referent (“illiterate”=0). There are historical factors related to the hmited access to higher education for the older participants in this study at the turn of the twentieth century, a period of colonial rule under the Japanese in 1895 (Tsurumi, 1977). It was the elite Taiwanese who either worked for the Japanese or those who could afford to send their children to higher R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 99 educational institutions that had attained education (Zimmer et al, 2003). The income variable was asked “How much total income did you and your spouse received last year?” with the respondent giving an answer in the New Taiwan (NT$) dollar amount. Income was divided into lower (less than NT$ 120,000), median (NT$ 120,000 to $599,999), and upper quartiles ($600,000 and above). It is important to note that the elderly who earn US$14,285 or less are eligible for a national elderly pension program that entitles them to US$85, which is underutilized by the population due to the welfare stigma associated with it. Among the respondents in this sample, only 1.7% would qualify for the program, which is .5% lower than the national estimated number of older adults who are eligible. Sociodemographic characteristics of children. The age of the children are asked in years and recoded into four age groups with the reference group as those who are in the adolescent years inclusive of their first year of adulthood (ages 12 through 18). The next three age groups are coded as separate dummy variables of interest: those who are “ages 19 through 24” (=1), those who are “ages 35 through 54” (=1), and those who are “ages 55 and over” (=1). The birth order of coresident child is an important factor to consider because the oldest child, especially the son, is expected to care for the older parents in Confucian value, according to Agree and colleagues (1998). In the survey, the interviewers were asked to “specify relation, birth order (e.g., eldest son, second son. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 100 eldest daughter, second daughter, etc.).” This variable is recoded with the only child as the reference group (“only child”=0) and the eldest child is coded as the variable of interest (“oldesf-1). In addition, the youngest and in-between children are also variables of interest (“youngest’-1 and “in-between”=l, in respective order). The differences between sons and daughters are also important to consider as their role expectations that accompany each gender. Consistent with the coding used for older parents, “females” (=1) are the group of interest and “males” (=0) belong in the referent group. Among older respondents who are living with at least one child, only 5% live with a daughter, while the majority live with their sons and daughter-in- laws. The place of residence of the children in relation to the older adult is important in the study because coresidence has been an important resource for support to be given and received among generations. In this survey, interviewers were directed to first ask about children who “live together with respondent,” and those who “live away from home.” Those who belong in the former are recoded as the reference group (=0). Among the children who “live away from home,” respondents were asked whether or not they “lived next door” (=1), in the “same neighborhood” (=1), lives “elsewhere in Taiwan” (=1), and “lives in China or overseas” (=1). There are one-half of the older adults in the sample who are living with married sons and 45.7% live with a daughter-in-law. The educational backgroimd of the children are asked individually for those who are six years or older, the years of education completed, which are divided R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 101 into three groups with those who have graduated from “elementary school or less” as the reference group (=0). The groups of interest are coded separately as those who have attained “some high school education or are high school graduates” (=1) and those with “some college or more” (=1). The average level of education attained by coresident children is 10.4 years with a standard deviation of 3.7. For those who have noncoresident children, the average level of education attained is 11.8 years with a standard deviation of 3.7, which is significantly an average of 1.16 years higher than the coresident children’s education in a paired samples two-tailed t-test at the p<.001 level. The occupational status of the children who are six years or older are asked if they are “working” (=1), a “student” (=2), “military” (=3), “housewife” (=4), “not working, unemployed, retired” (=5), “part time work/study” (=6), or “other (please specify)” (=7). Those who are “students, housewives, not working, unemployed, retired, or other” are coded as the reference group (unemployed=0). Those who are working “full time” are a group of interest (“fiilltime”=l). In addition those who are working “part time” and those who are working for the military are separately coded as a dummy variable, “part time” (=1). The marital status of each adult child are asked only those 15 or older as an ordinal variable, coded with the first category as “married” (=1), “cohabit” (=2), “separate” (=3), “divorced” (=4), “widowed” (=5), and “unmarried” (=6). Those who belong in categories one through two are coded as “married or living with a partner” (=1) and the rest of the categories of being “single, divorced, or R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 102 widowed” belong in the reference group (=0). Among the older Taiwanese who coresides with at least one child, 72.9% of these coresident children are married. Respondents were asked the number of grandchildren that they have for each adult child above 15 years of age, which was recoded as a dichotomous variable with those who have no children as the reference group and those who have at least one grandchild as the group of interest. There are 53.4% of the total respondents who live with at least one grandchild who is under 18. The recoded variable used in this dissertation for frequency of contact between the adult child and the older adult is a combination of the two questions: “How often do you see him/her?” and “How often do you talk on the phone?” Those who “hve with their parent” are coded as the reference group (=0). There are four other groups of interest, those who are living apart from the older respondent, but see or “talk to each other everyday” (“daily”=l), “every few days to weekly” (“weekly’- 1), “every few months to monthly” (“monthly”=l), “every few years or yearly” (“years”=l). Overview of Statistical Analysis Since this dissertation is proposing that intergenerational family support exchanges can be characterized as a circumscribed set of "ideal" types that are empirically manifested by combinations of observed variables, latent class analysis (LCA) is used to examine the typological structure underlying intergenerational solidarity. The advantage of using LCA over traditional types of R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 103 cluster analysis includes probability-based classification. Cases are classified into clusters based upon membership probabilities estimated directly from the model. This dissertation will use Latent Gold Software 2.0 (Vermunt & Magidson, 2000). In addition, the multinomial logistic regression is performed to predict the type of support patterns that occur between generations based upon characteristics possessed by the older respondent and their children under each model of motivation. The software used is ST AT A 7 (StataCorp, 2001). Latent Class Cluster Analysis What is Latent Class Cluster Analysis? LCA is a statistical method that allows researchers to uncover the latent variables, which are the unobserved groups in which persons (cases) fit into who share similar interests, values, characteristics, or behavior beyond direct causal relationships and cannot be directly measured (Clogg and Goodman 1984; Lazarsfeld and Henry 1968; McCutcheon 1987). A key assumption of LCA is that membership in a latent class is the true source of covariation among measured categorical variables. Thus, a given set of latent classes is acceptable to the extent that it minimizes the within-class cross-classification among observed indicators— the assumption of local or conditional independence. This property underlies a statistical test of whether a theoretical model adequately describes the observed data and provides a basis for comparing alternative theoretical specifications. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 104 There are two types of variables used in latent class analysis. The first are static latent variables, which are either expected to be constant over time, or the change is not of interest in a particular study, as when group differences at a particular time are examined. The second type is the dynamic latent variable, which are used when looking at change in systematic and important ways over time. Of ten the same variable may be thought of as static in one context and dynamic in another, depending on the objectives and interests of particular research project (Collins & Flaherty, 2002). Indicators are the variables that are used to define or measure the latent variables. Indicators maybe treated as nominal, ordinal, continuous, or count. Covariates are the variables used to describe or predict the latent variable may be treated as nominal or numeric. The estimation of parameters for LC cluster models involves both the maximum-likelihood (ML) and the Posterior Mode (PM) methods derived from the probability density function defining the models. The ML method allows LC analysis to estimate the model parameters. While the PM allows classification of objects into clusters based on the posterior class-membership probabilities, using the standard classification method of modal allocation (assigning each object to the class with the highest posterior probability). Formally, the latent class model with latent variable X with T categories (corresponding to latent types of intergenerational relations) which are described by variables A, B, and C whose levels are indexed by ij,k respectively. The probability of membership in an observed cell is defined as follows: R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 105 TUijk - ' ^ J t x (t) n:Ajc=t (i)7 C B j:= t (j) 7 t c x = t ( k ) where iixit) is the probability thatX= t, n.Ax=t (0 is the conditional probability that item A takes on level i, given that latent variable X is at level t; the other conditional probabilities are defined similarly. The expectation-maximization (EM) algorithm is used to begin the estimation with a number of iterations. When the estimation is getting close to the optimum solution of clusters, the program then switches to the Newton-Raphson algorithm. Latent Gold 2.0 resolves the occurrence of local solutions by using multiple sets of random starting values. Model selection is based upon standard likelihood-ratio tests between models using various statistical tests to determine the goodness-of-fit for cluster models with the data, which are further discussed under the sub-section, classification and measurement, within this section. The standard latent class model is a model for measuring one or more latent (unobserved) categorical variables by means of a set of observed categorical variables; these observed variables are considered to be indicators of the underlying concepts. The model is discussed in terms of two closely related parameterizations: (1) a parameterization in terms of probabihties of belonging to a particular latent class and of obtaining a particular scoring pattern on the observed variables, given the latent class one belongs to; and (2) a parameterization in terms of a loglinear model with a categorical latent variable. Since research problems often require R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 106 particularly restricted models (e.g., the probably of giving a "wrong" response is the same for all indicators, or the relationship between the latent variable and the indicators is linear), attention is paid to imposing such restrictions by use of either parameterization. Finally, a multigroup or comparative latent class analysis will be used, a technique that is useful to researchers who want compare the outcomes of a latent class analysis for samples independently drawn from the same population but at different points in time. Classification and measurement. The latent variables are treated as theoretical constructs and the observed variables as indicators of these constructs. They focus on the question of whether the (many) observed categories of one indicator or the observed scores on a number of indicators may be reduced to a small number of fundamental, theoretically meaningful latent categories or classes. The underlying classes may be treated as a classification, that is, as categories of nominal level latent variables, or they can be regarded as ordered, or as categories of interval level latent variables. In this dissertation, the categorical latent variables are regarded as nominal level variables, whose categories are not ordered; hence, there is a close connection between the concepts of "cluster" and "latent class." Latent class cluster analysis is essentially a variant of what Gibson and Lazarsfeld in the 1950s called latent profile analysis, in which the underlying variable is treated as a nominal level latent variable, but the observed variables (the indicators) are treated as continuous (unlike the standard R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 107 latent class model with categorical indicators). Latent class cluster analysis has been used in a variety of disciplines from psychology to clinical medicine. The advantage of using latent class cluster analysis over correlational analyses (e.g., regression, factor analysis) is the ability for researchers to move beyond focusing on how similar respondents are to each other; thus, allowing to capture the mix of relationships that exist within the group that otherwise would be considered “in error” and representing random associations. As researchers in the field of gerontology that values interdisciplinary understanding of diversity within research, gerontologists can benefit from expanding their armamentarium of statistical approaches that enhance the likelihood of uncovering diversity. Cluster analysis indeed is one such approach, identifying subgroups of people who are similar to each other on various characteristics (Aldenderfer & Blashfiedl, 1984). When a researcher has only categorical data and cannot assume, for theoretical or methodological reasons, that the observed variable are discrete representations of an underlying continuous variable, latent class analysis (LCA) is another tool for identifying taxonomies or meaningfully different subgroups within a research sample. While there has been little use of LCA in recent gerontological literature, LCA has been used in many ways in other fields, including exploring statistical techniques, developing psychometric models, and examining them in substantive ways. An example of the first was Goodman’s demonstration of the use of LCA in R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 108 path analyses (1973a, 1973b). Several authors have used LCA to develop psychometric and design models. These include comparisons among different groups on scale response or typology (Birkelund, Goodman, & Rose, 1996; Clogg & Goodman, 1984, 1985, 1986), characterizations of responses to attitude questions (Brody, 1986), and modeling item-choice behavior (Shigemasu & Sugiyama, 1984). LCA has been used for substantive issues such as understanding comorbities of adolescent problem behaviors (Fergusson, Horwood, & Lynskey, 1994) and subtyping schizophrenia (Castle, Sham, Wessely, & Murray, 1994). Introducing the more complicated latent class cluster models for “mixed variables,” that is, models in which some indicators are continuous but others are categorical, possibly with ordered categories, extends the latent profile model here. In addition, according to the method suggested by Vermunt and Magidson (2002), covariates are introduced into the latent class cluster model to find the "causes" of belonging to a particular cluster rather than to another. In this analysis, LatentGold software (Vermunt & Magidson, 2002) will be used to perform latent class cluster analysis based upon the support exchange items. From the literature, a four-class model is hypothesized of social support transfer models, which include upward transfers, reciprocal exchange, downward transfers, and none. According to Plunkett and Mitchell (2000), one technique in fitting a theoretically and empirically meaningful model is to start with a two-class model; if that model does not fit, one can then move on to a three-class model and so on until a goodness-of-fit criteria is reached using the likelihood ratio chi-squares (L^), degrees R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 109 of freedom, Bayesian Information Criterion (BIG), and significance levels for the five models under consideration that are listed in Table 7 in the next chapter. The tests for statistically significant discrepancies betvv'een a theoretical model and the observed data, providing a basis forjudging the adequacy of a given specification through statistical inference. The BIG statistic (Raftery 1986) is useful when selecting the best fitting model among competing models, especially when choosing among non-nested models and where large sample size causes otherwise acceptable models to be rejected based on the \}. The BIG is calculated with the following equation: - [{df)\n{n)], where n is the sample size (here, 9477). The most desirable property of the BIG is that, compared to the L ", it is less likely to disadvantage more parsimonious models — those that have fewer latent classes and estimate fewer parameters — in the model selection process (Glogg 1995). From this information, preliminary decisions about the comparison of models and model selection are made. Causal analysis and dynamic models Once a theoretically and empirically reliable latent typology or latent scale has been found, researchers often seek to investigate the causes and consequences of the latent variables, along with the way the several latent variables R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 110 influence each other or develop over time. Dayton and Macready (2002) extend the basic latent class model by introducing independent variables that influence (or predict) the probabilities of belonging to particular latent classes. If the independent variables are categorical, the resulting latent class model is designated as a model with blocking variables. As the authors show, this "blocking" model is a straightforward extension of the multiple-group latent class model. Formulated in a "causal" language, the test of homogeneity is measured by whether the associations between the categorical covariates and the indicators are completely mediated by the latent variable (the homogenous case) or whether there remains a direct effect of the covariates on the indicators after the latent variable has been controlled for (the heterogeneous case). Along with categorical covariates, the authors show how to introduce continuous covariates into the standard latent class model. Because the dependent (latent) variable is still treated as categorical, the relationship between the continuous independent variables and the latent variable is modeled in the same way as in the logistic regression model. Two kinds of parameters are estimated for each model tested: conditional probabilities and latent class probabilities. Conditional probabilities reflect the distribution of observed indicators for members of each latent class. These estimates are analogous to factor loadings in that they represent the association between observed and latent variables, and are useful for characterizing the nature of the latent classes. Latent class probabilities signify the distribution of R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. I ll members across types, making it useful for describing the prevalence of types within a population and for comparing prevalence between populations. Multinomial Logistic Regression In behavioral and social data, it is common to have a nested structure, as in this study; dyads of older respondents are matched to each of their children on the household roster. Since it is possible that there are both variance and covariance components within the nested structure (e.g., the correlation among individual-level variables and within- and between-family components), improved estimation of effects within individuals and families, and across individuals and families must be accounted for in the analysis. Thus, STATA is one such program that can predict cluster membership while controlling for variables that possibly differ within and across individuals and families in the dataset. Thus, given the nested structure of the data and the discrete dependent variables, the multinomial logistic regression in STATA is a logical choice for this dissertation. The mlogit command in STATA stands for maximum likelihood multinomial (polytomous) logistic regression, which performs maximum hkelihood estimation of models with discrete dependent variables (StataCorp, 2001). It is intended for use when the dependent variable takes on more than two outcomes and the outcomes have no natural ordering. In this dissertation, there are four possible outcomes of emotional and instrumental upward transfers, no upward transfers, economic upward transfers, and full upward transfers. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 112 The default in STATA is to use the largest group as the baseline comparison group for the outcome. However, in this dissertation the no upward transfer type is the natural choice that will be used as the baseline group to which the other three upward transfer types are being compared. The models of motivation are run in a hierarchical manner within the multinomial logistic regression (Le, 1998). Each model is represented by the relationships among its variables, and specifies how variables added to the first model influence relations occurring in the subsequent models. As mentioned above, there are four models that are tested in the following order: need, exchange, resources of power, and ideology. Beginning with Model 1, which is the need domain or altruism model, the older parent is expected to receive transfers from their adult children, which are tested by using the support provision variables from the older adult to the adult children as predictors for the no upward transfers pattern, as compared to each of the other three upward transfer patterns. The next model is the exchange domain (Model 2), which includes variables representing short-term and long-term support given from the older adult to their adult children as predictors for upward transfer patterns, representing reciprocity. Independent variables representing Model 3 is added to the previous model in order to test for resources of power that the older person has to command the type of support that is manifested as the outcome when compared to no upward transfers. The fourth model tests the level R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 113 of filial norms expected by the older adult in predicting the ratio of each type of upward transfer as compared to the baseline support category of no upward transfers. The models of motivation are inferred from the variables in four ways. The first is when one or more independent variables are only significant in one model. For example, if older parents who are in poor health significantly predicts receiving emotional and instrumental health from the adult children only in Model 1, then it can be inferred that the estrangement model is partly explained by this specific relationship between giving and receiving support. The second inference is drawn fi’ om the significant independent variables that remain significant in two or more consecutive models and continues to remain significant in the final model, which helps explain why the final model predicts the corresponding type of upward transfer pattern, if any. Another way of inferring a model of motivation is by examining the significant variables that remain significant in two or more consecutive models, but do not remain significant in the final model. In this case, the predictors support the last model of motivation in which it was significant. The fourth method of inferring a model of motivation is found when the variables lose significance in one model, but become significant again in a later model. Just as in the third inference method, the significant variables are inferred to explain the relationships between the model of motivation and the corresponding transfer pattems that they last remained significant after adding the subsequent models. For example, if in Model 2, only the economic need variable represented by enrollment R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 114 in welfare significantly predicts economic upward transfers, and these variables were not significant in all the other models of motivation, then this variable operationalizes the altruism model based upon economic need. The fourth statistical model, which is the ideology domain, is the main model of interest. The filial normative levels of expectations are included to see if there is any significant differences explained by the older person’s perceptions of filial piety that are manifested in the social support pattems. By adding in variables to build upon previous models, this will allow the researcher to examine the effects of the relationship between the motivations and the support pattems by taking into consideration overlapping characteristics of the older parents and their adult children that are possibly present across all models. The next chapter will go over the results found in carrying out LCA and the multinomial logistic regression, using Latent Gold and STATA softwares, in respective order. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 115 CHAPTER FIVE Results In this chapter, the results for the latent class cluster analysis and the multinomial logistic regression are analyzed in order to answer the central question of this dissertation: how are filial expectations related to social support exchange pattems between older Taiwanese adults and their adult children? First, the results of the latent class cluster analysis (LCA) are presented to determine the types of social support pattems that exist in the Taiwanese population between older adults and their adult children. Second, multinomial logistic regression is tested as hierarchical models of four domains of motivations as specified in chapter four. Based upon the latent clusters found in LCA, this second section is further divided into three comparisons of outcomes: 1) emotional and instmmental upward transfer class versus no upward transfer class, 2) economic upward transfer versus no upward transfer class, and 3) full upward transfer class versus no upward transfer class. Forming Latent Class Clusters for Directions of Social Support Transfers This section will describe how results are obtained for 1) the selection of indicators and 2) determining the best-fit model using LCA. Characteristics of the samples are described in Table 5 of the previous ehapter, which shows the variation in the distribution of all variables. Due to the sparseness of the data for older parents R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 116 giving daily assistance through babysitting their grandchildren (0.2%), giving material belongings (0.2%), giving money (2.0%), and receiving daily assistance (2.0%), these three indicators are omitted in LCA, but the The determination of the four-class model is based upon a theoretically and empirically meaningful model. In terms of a theoretically sound model, chapter 3 describes four types of classes found in the literature: upward, downward, reciprocal exchange, and none. However, due to the sparseness of the variables operationalizing the support given by the older parent to their adult children, the downward, reciprocal exchange, and no transfers cannot be estimated in LCA. Hence, within the upward transfer class, the following are eight possible sub classes of transfers that can be found through this exploratory analysis: emotional upward transfers, instrumental upward transfers, economic upward transfers, emotional and instrumental upward transfers, emotional and economic upward transfers, economic and instrumental upward transfers, full upward transfers across all items, and no upward transfers. Still staying within the theoretical framework, the number of possible classes is in part determined mathematically by the three upward types of emotional, instrumental, and economic support. The following is a simple algebraic equation determining the maximum possible pattems of support: Sub-types of Upward Transfers max-n^ -1 with « = 3, representing the three types of upward transfers. As for the empirical determination of the best-fitting model, the method is based upon testing a series of latent class stractures, successively adding R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 117 an additional class and observing the following goodness-of-fit measures: the likelihood ratio chi-square test statistic (L^), the Bayesian Information Criterion (BIG) statistic, and the probability of obtaining values of the test statistic that are equal to or greater in magnitude than the observed test statistic (p-value). Table 7 shows the L ^, BIG, df, and the p-value for each model. The first class is the independence model, which is always the worst fit (L^=l5996.82, BIG=15758.75, df=26, and p<.001) (Table 7). When examining the rest of the models, the four-class model has the lowest BIG of -67.61 and a non-significant p-value (p>.100). But, in order to ensure that the four-class model was the best fit, an extra class was tested to see if the lowest BIG value was reached in a four-class versus five-class model. With the addition of each model, the BIG value shows a precipitous drop with the introduction of each additional class up to the four-class model, at which point it appears to reach an asymptote and little further improvements are attained (Figure 2). The goodness-of-fit statistics indicate that the likelihood ratio was not statistically significant (L ^ =5.65, df=8.00, p>.100), revealing a relatively good fit of this model to the data (Table 7). Thus, the four-class model was the preferred choice based upon both the predicted theoretical framework and empirical goodness-of-fit measures. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 118 Table 7. Likelihood Ratio Chi-Squares for Selected Models. Model Description LHorLL) BIC df (or Npar) p-value 1 One-Class 15996.82 15758.75 26.00 3.10 X 10-'^ “ 2 Two-Class 469.77 286.64 20.00 0.00 3 Three-Class 232.62 104.43 14.00 0.00 4 Four-Class 5.65 -67.61 8.00 0.69 5 Five-Class 2.03 -16.29 2.00 0.36 Figure 2. Model Fit Evaluated by BIC for Classes 1-5 350.00 300.00 - 250.00 200.00 y 150.00 - “ 100.00 ^ 50.00 ^ 0.00 - -50.00 - - 100.00 1 2 3 4 0 5 6 Number of Classes The conditional latent class probabilities for the four-class constrained model are reported in Table 9. The relative sizes of these probabilities describe the profile of each class. The first class has a probability score of 34%, which indicates that the majority of older adults in Taiwan have a higher probability R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 119 of receiving both emotionally and instrumentally, suggesting an emotional and instrumental upward transfer class (Figure 3). The second type reveals lower scores on receiving emotional, instrumental, and economic support, when compared to the rest of the classes, suggesting no upward transfer type of intergenerational support, which amounts to a one-third of the older population in Taiwan who are likely to fall into this category (Figure 4). The third type is characterized by high material and monetary support probabilities, labeled as economic upward transfer with a probability of 30% of the older population who fit in this category (Figure 5). Finally, the fourth type has high probabilities of receiving across all measures of support, suggesting the full upward transfer type of relationship, which possibly represents only 3% of the older Taiwanese population (Figure 6). Table 8. Goodness-of-fit statistics for the Four-Class Model 4-Cluster Model Coefficients Number of cases 9477.00 Number of parameters (Npar) 23.00 Random Seed 34126.00 Chi-squared Statistics Degrees of freedom (df) 8.00 p-value L-squared (L^) 5.65 0.69 X-squared 4.74 0.78 Cressie-Read 4.91 0.77 BIC (based on L ^) -67.61 AIC (based on U-) -10.35 CAIC (based on L ^ ) -75.61 Log-likelihood Statistics Log-likelihood (LL) -17765.15 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 120 Table 8. Goodness-of-fit statistics for the Four-Class Model (continued) Log-prior -13.32 Log-posterior -17778.47 BIC (based on LL) 35740.91 AIC (based on LL) 35576.31 CAIC (based on LL) 35763.91 Classification Statistics Classification errors Reduction errors Entropy R-squared Classification log-likelihood AWE 0.14 0.78 0.73 -20837.44 42165.08 Table 9. Conditional Latent Class Probabilities in a Four-Class Model Emotional and Economic Full Instrumental No Upward Upward Upward Upward Transfer Transfer Transfer Transfer Class Class Class Class 0.34 0.33 Overall Probability Indicators Support Received from Parent to Adult Children 0.30 0.03 Receive Love and Care 0.99 0.00 0.00 0.98 Receive Listening 0.59 0.11 0.06 0.69 Receive Material Belongings 0.03 0.00 0.24 0.54 Receive Money 0.09 0.01 0.46 0.84 Receive Care when 1 1 1 0.94 0.00 0.00 0.92 The body of Table 9 contains the (partial) conditional probabilities that show how the clusters are related to the indicator variables, specified as nominal, R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 121 with “yes”=l and “no’-O. Within each class, the probability coefficients arc listed for those who give or do not give the support for each indicator, which sum up to 1, only those who answered “yes” are shown in Table 9. The emotional and instrumental transfer class shows that among the older adults who fit in this category, there is a 99% probability that the older adult feels loved and cared for by his or her adult child and 59% are likely to have an adult child that listens to the older parent’s worries and concerns. At the same time, older parents are 94% likely to receive care from their adult children when they are ill. Hence, the emotional and instrumental care items are highly likely to be given to the older adult simultaneously. In the no upward transfer class, the probability for any transfers from the adult child to the older parent is close to 0%, except for 11% who listen to the older adult talk about his or her worries and concerns. When examining the characteristics of the older respondents within the economic upward transfers class, the probability for parents to receive economic support is highest in this class relative to the other classes of support, with 24% of the older respondents likely to receive material belongings and 46% money. The full upward transfer class consist mainly of older people whom have children willing give across all types of transfers (92% love and care, and 64 % listening), instrumental (98% care when ill), and economic support (54% material and 84% money). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 122 Rgure 3. Errctional and fnstrymental Upward Trarffifer Q ass (N=3362) m m O t; w C 0) e e = -J i3 -= ™ C 5 ^ 5 o o ^ ■ J 3 0 . c o o 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 Receive Love Receive Listening Receive Material Receive Money Receive Care and Care Belongings when III Support Indicators R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 123 Figure 4. No Upward Transfer Class (N=4269) a (Q £! O m w a O c S 2 a c o § c o o 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 Receive Love Receive Listening Receive Material Receive IVtaney Receive Care and Care Belongings w hen I I I Support Indicators Figure 5. Economic Upward Transfer Class (N=1720) R eceive Love and R eceive Listening R eceive M aleriai R eceive M oney Receive C are when C are Belongings iii Support Indicators R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 124 1.00 T Figure 6. Full Upward Transfer (N=127) I R eceive Love and R eceive Listening R eceive Material R eceive M oney R eceive C are w hen C are B elongings I II Support Indicators Multinomial Logistic Regression Using this typology as the dependent variable, a multinomial logistic regression analysis is performed to test the models of motivation as predictors for the pattems of upward transfers. Specifically, maximum-likelihood multinomial (polytomous) logistic regression is used to specify that the observations are independent across and within groups or clusters. As a default function in STATA, the baseline cluster chosen as the comparison group is the largest group, which in this dissertation is the no upward transfer group. The other three types of exchanges will he compared against the type of exchange pattem that occurs in the majority of the population. The complete results from the multinomial logistic regression are found in Tables 10, 11, and 12 (Appendix A), but in order to simplify the results to R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 125 the important findings, only the motivational factors are found in the text. Each table is presented as the individual-level factors associated with the following three classes of intergenerational support as compared to the no upward transfer class: emotional and instrumental upward transfer, economic upward transfer, and full upward transfer. The analysis begins with the independent variables operationalizing sociodemographic characteristics of the respondents and their children, and the types of support that the older adult provides for each adult child. As described in the previous chapter, the models of motivation are presented as four domains: need (altruism), exchange (mutual aid), resources of power (power and bargaining), and ideology (filial norms). Upon examining the pseudo R^, the percent of variance explained increases across the models from 8.91% in Model 1 to 9.60% in Model 4. The greatest change in the percent of variance explained, which is 0.70%, occurs between Model 2 and Model 3 with the addition of the variables operationalizing the resources of power domain where parents provide their own resources for the children through division of property, corporate control (as the main economic decision maker in the household), and the income generator (as the main wage earner in the household). In terms of the -2 log likelihood, the coefficient decreases across the models, with the greatest decline of 166.80 in the -2 log likelihood occurring between Model 2 and Model 3. This indicates that this measure is an independently significant factor contributing to the probability of an older adult engaging in upward transfers rather than no upward transfers (A X^=30520.44, Adf = 144, p<.001). Thus, both the percent of variance and the log likelihood estimates show that adding in the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 126 subsequent variables improves the previous models, especially when adding in the resources of power variables, which produces the greatest increase in the percent of variance explained. However, this percent increase is less than one percent, which is a very small increase in the percent of variance explained. Emotional and instrumental upward transfer class versus no upward transfer class The odds ratios are presented in Table 10 for those who are engaged in the emotional and instrumental upward transfer class (with children giving to parents in the form of instrumental and emotional care) compared to those in the no upward transfer group (older parents neither receive emotional, instrumental, nor economic support from their adult children). The analysis begins with estimating the sociodemographic characteristics of the respondents and their children, and the types of support that the older adult provides for each adult child. Model 1 shows that the older adult’s age, gender, marital status, ethnicity, and place of residence are statistically significant. Those are in the old-old age group (75-84 years old) are 45% (p<.05) more likely than the oldest-old group (85 years and older) to receive emotional and instrumental upward transfers than no transfers at all. In addition, those who are single, female, of Holo and Chinese descent, and living in rural areas are more likely than their counterparts who are married, male, of Indigenous descent, and living in rural areas, to receive emotional and instrumental care from their married daughters who have at least a college education and an only child. Further, adult children who are in older age groups. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 127 when compared to those in the younger age groups, are 44%-58% (p<.001) less likely to give emotionally or instrumentally to their older parents, than not giving at all. The birth order of the adult child is also significant, but only for those who are the youngest child and who are one of the middle children, when compared to the only child. The older parent who has a youngest child has less than a 25% (p<.01) likelihood of receiving emotional or instrumental help when compared to the only child; while those who have an in-between child of three or more children, are 50% (p<.001) less likely than those with an only child, to receive emotional or instrumental help rather than no help. Contrary to expectations, the need variables introduced in Model 1 do not show any significance in determining emotional and instrumental upward transfers when compared to no upward transfers. In Model 2, the addition of the exchange variables demonstrate that older parents who give immediate care or have made past investments are less likely receivers of support. Even though there are changes in the significance levels for the child’s gender, education, and martial status, upon the adding the mutual aid variables, the odds ratios for the characteristics of the children who are giving emotional and instrumental care to older parents rather than no care at all remains relatively the same throughout the rest of the models. The next model illustrates that older people with resources of power are likely receivers of emotional and instrumental support from their children. When compared to those who have no property to divide, those who have property and have either partially divided it or not yet divided it have an odds ratio of 1.37 (p<.01) R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 128 and 1.24 (p<.01), respectively. Those who have corporate control or who are main economic decision makers in the household are 16% (p<.05) more likely, than those who are not, to be receivers of emotional and instrumental care rather than no care at all. Furthermore, those who are the main income generators in their home are almost 40% (p<.001) more likely to than those who are not the main income generators, to receive emotional and instrumental care. As in the previous model, the sociodemographic variables of the older parents and the child, as well as the exchange variables, remain significant with relatively minor changes in the coefficients. The final model shows that ideology is a significant factor to consider for older Taiwanese adults who are likely receiving emotional and instrumental upward transfers rather than no transfers at all (odds ratio of 1.02, p<.01), but it is a very small increase of only 2% more likely with increasing levels of norms. The variables that were significant in Model 1 through 3 continue to remain significant in Model 4, with the largest odds ratio increase in marital status of the child from 20% (p<.05 in Model 1) to 44% (p<.001), which portrays the higher likelihood of older adults who have married children rather than unmarried children, offering emotional and instrumental help to them. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 129 Table 10. Simplified table of odds ratios from multinomial logistic regression of being in emotional and instrumental upward transfer class vs. no upward Variables Model 1 Model 2 Model 3 Model 4 Models of Motivation Need: Altruism Health Need: Self-rated health Good health 0.99 1.00 0.97 0.97 Average health 0.99 1.00 0.98 0.98 Poor health(reference) Economic Need: Welfare status Enrolled in elderly welfare 1.02 0.99 1.01 1.09 program Not enrolled in elderly welfare program (reference) Exchanges: Mutual Aid Immediate Babysit(No=reference) 0.53*** 0.52*** 0.53*** Money (No=reference) 0.58** 0.56** 0.56** Investment House, education abroad, or 0.45*** 0.42*** 0.42*** medical treatment No past investments (reference) Resources of Power. Power and Bargaining Property ownership No property to divide (reference) All property divided 1.20 1.21 Partially divided property 1.37** 1.38** Not yet divided property 1.24** 1.24** Corporate Control 1.16* 1.15* Main economic decision maker (No=reference) Income Generator Respondent is main wage eamer for household 1.36*** 1.35*** Respondent is no main wage earner for household (reference) Ideology: Filial Norms Levels of norms 1.02** *** p<.001, **p<.01, *p<.05 Economic Upward Transfer versus No upward transfer class Table 11 presents the odds ratios for models that predict the characteristics of older adults who will be in the economic upward transfer groups (older parents are receiving financial and material support from their adult children) versus those who are in the no upward transfer group. Looking first at Model 1, the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 130 older parent’s age and gender significantly predict the type of transfer class that they belong in. Those who are in the young-old age group (age 65 to 74 years old) are less likely than those who are the oldest old (age 85 and older) to be in the economic upward transfer class than the no upward transfer class. While female are 27% (p<.05) more likely than males to be in the former rather than the latter class of support. Furthermore, the child’s age group is a significant predictor in this model, when compared to those who are in the youngest age group (ages 12 through 18), those who are older in the following two age categories: ages 19 through 34 years old and 35 through 54 years old, are less likely givers of economic upward transfers to older parents. Upon examining the need variables under the altruism model, older Taiwanese adults who are in good health are 35% (p<.01) less likely that those who are in poor health, to become receivers of economic help, rather than no help, from their children. Thus, poor health is a proxy for an older person’s need for economic support. The addition of the variables operationalizing the exchange domain in Model 2, alludes to the prediction that older adults in good health who have been offering support are less likely going to be receivers of economic aid from their children. The sociodemographic variables in the previous model remain significant with the addition of the child who works frill time is 1.22 (p<.05) times more likely than the unemployed child to offer economic support to their parent. In Model 3, controlling for resources of power illustrate that older adults who have partially divided property that they have are 53% more likely R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 131 (p<.05) than those who have no property to receive monetary and material help from their adult children. In addition, older adults who are the main wage eamer for the household are 37% less likely (p<.01) than those who have children as the main wage eamer, to receive financial and material support rather than no support at all from their children. As in Model 2, the sociodemographic variables remain significant with relatively small changes in the coefficients. The final model shows that the levels of filial norms perceived by the older Taiwanese parent does not make a difference in determining receiving monetary and material support as compared to receiving no support. Again, the variables that were significant in the previous model remain significant. Those who are in good health have an odds ratio of 0.66, which shows that respondents with good health when compared to those with poor health are less likely going to receive monetary or material transfers from their children. Table 11. Simplified table of odds ratios from multinomial logistic regression of being in economic upward transfer vs. no upward transfer (comparison group) for models of motivation (N=9477) Variables____________________ Model 1 Model 2 Model 3 Model 4 Models of Motivation Need: Altruism Health Need: Self-rated health Good health 0.65* 0.66** 0.67** 0.67** Average health 0.80 0.81 0.81 0.81 Poor health(reference) Economic Need: Welfare status Enrolled in elderly welfare 1.04 1.02 1.01 1.01 program Not enrolled in elderly welfare program (reference) Exchange: Mutual Aid Immediate Babysit(No=reference) 0.54J 0.53* * 0.53*J Money (No=reference) 0.45* 0.44* 0.44* R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 132 Table 11. Simplified table of odds ratios from multinomial logistic regression of being in economic upward transfer vs. no upward transfer (comparison group) for models of motivation (N=9477) (continued) Variables Model 1 Model 2 Model 3 Model 4 Investment House, education abroad, or 0.37*" 0.35*” 0.35*” medical treatment No past investments (reference) Resources: Power and Bargaining Property ownership No property to divide (reference) All property divided 1.25^ 1.25^ Partialiy divided property 1.53* 1.53* Not vet divided property 1.20 1.20 Corporate Control Main economic decision maker 1.28 1.28 (No=reference) income Generator Respondent is main wage earner for household 0.63 * 0.63*” Respondent is no main wage earner for household (reference) ideoioav: Filial Norms Levels of norms____________________________________________________ 1.00 *** p<J01, **p<.01, *p<.05 Unlike the results in Table 9, the older person’s health does have a significant effect upon the pattems of support between the older parent and his or her child, whereas the elderly perspective of filial norms does not have any effect. Another difference between the two tables is found in the fewer number of sociodemographic variables that predict receiving economic support rather than no support across models. Full upward transfers versus no upward transfer class When testing the predictors for older adults who fit in the traditional model of receiving transfers aeross all types of support from their adult children, as compared to those who are in a nontraditional model of reciprocity, the older adult’s income level and frequency of intergenerational contact becomes significant R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 133 predictors for the first time. Table 12 shows that those who are single, living in rural areas, who have an income in the median quartile range, are more likely than their counterparts to receive economic, emotional, and instrumental support from their adult children (p<.01). In addition, the characteristics of the children who are more likely to give all of these support items as opposed to not giving any support include those who are working daughters who see their parents daily, but not in coresidence. In Model 1, older adults who are unmarried, living in rural areas, and have an income level in the median quartile range, when compared to those who are married, living in urban areas, and who have income in the lower quartile range, are more likely receivers of full upward transfers of care than no upward transfers. In terms of the children’s characteristics, adult children who are working part time when compared to the unemployed group, are 61.38% (p<.01) more likely to provide upward transfers to their older parents, which is an overwhelming odds ratio that is detected in skewed data through the use of multinomial logistic regression. Those who have daily face-to-face contact, but are in noncoresidence with their adult children, are more than eight times (p<.05) as likely as those who are have daily face-to-face contact in coresidence to receive emotional, instrumental, and economic support than no support at all. In the next model, the inclusion of past investments made by older adults in their children’s house, education, or health shows that they have an odds ratio of 0.20 (p<.001), which shows that these parents are less likely receiving any social support rather than no support. The significant characteristics of the older R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 134 adults and their children from Model 1 remain significant in this model. Therefore, Model 2 shows that older parents who have provided support to working daughters who they see daily, but not in coresidence, are less likely receiving full upward transfers and no upward transfers. In Model 3, the addition of variables operationalizing the concept of power and bargaining, shows that older parents who are the main economic decision makers in the household is more than twice as likely than those who are not to receive full upward transfers of support from their adult children. As in the previous model, all of the significant variables remain significant, with the addition of gender and age group that the older parent belongs to. Those who are in the young-old age group are more than 60% (p<.01) less likely than those who are at least 85 years old to receive full transfers of support, rather than no transfers from their children. Older females are almost twice as likely than males to receive full upward transfers of support from their children, rather than no transfers. Among the child’s sociodemographic characteristics that become significant in this model is the child’s marital status, as those who have adult children that are married or living with a partner are almost twice as likely as their counterparts to receive full upward transfers rather than no upward transfers. In the final model, the ideology of the older Taiwanese parent does not matter in predicting full versus no upward transfers of support, as the filial norms model is not significant. The sociodemographic variables operationalizing the background characteristics of the older parents and their adult children that are R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 135 Table 12. Simplified table of odds ratios from multinomial logistic regression of being in full upward transfers vs. no upward transfer (comparison group) for models of motivation (N=9,477) Precedent Mutual Aid House, education abroad, or 0.20*** 0.44*** 0.35*** medical treatment No past investments (reference) Resources Property ownership No property to divide (reference) All property divided 0.88 0.89 Partially divided property 1.40 1.40 Not yet divided property 1.20 1.29 Corporate Control Main economic decision maker 2.26** 2.24' (No=reference) Income Generator Respondent Is main wage earner for household 0.92 0.92 Respondent is no main wage earner for household (reference) Ideoloav: Filial Norms Levels of norms 1.02 Not enrolled in elderly welfare program (reference) Exchange: Mutual Aid Immediate Babysit(No=reference) 0.45 0.41 0.42 Money (No=reference) 0.86 0.84 0.83 *** p<.001, **p<.01, *p<.OS significant in Models 3 are also significant in this model. In terms of the filial expectations model, it is only significant when predicting emotional and upward transfers as compared to the no upward transfers, as seen in Table 9. A more thorough summary and interpretation of each of these models are given in the next chapter. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 136 CHAPTER SIX Discussion and Conclusions Contrary to popular beUef that older persons in Confucian cultures continue to rely heavily upon their children for emotional, instrumental, and economic support due to high filial normative expectations, this study suggests those older parents who have both economic power and have higher levels of filial normative expectations will receive only emotional and instrumental support from an adult child. Consistent with the modified modernization framework introduced in this dissertation, there is a gradual shift in intergenerational support patterns as receiving no upward transfers is the most prevalent type in Taiwanese society. These non-receivers are typically economically independent older parents and are providing immediate mutual aid and past investments in the children. These results can be generalized to the national population of Taiwan, shedding light on several important issues surrounding the rapidly changing social, economic, and demographics of the country that shows older adults as economically independent individuals who are providing support to their offspring. Unfortunately, the results are based only upon the older person’s responses about the types of support given and received between the generations and their own perceptions of filial norms, rather than on the responses from the younger generations as well. This one-dimensional response is a potential problem as there may be a bias in the older person’s response toward more positive portrayal than the actual circumstances, as R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 137 “saving face” or protecting one’s reputation is important in Asian culture. However, Hermalin and colleagues (1992) have noted that the majority of responses given by older parents were similar to their children, according to a sub-sample of the original respondent’s children collected during the second wave in 1993. The discussion is organized around the hypotheses presented in chapter three and the seven major models of motivations for intergenerational exchange: estrangement, altruism, corporate, power and bargaining, mutual aid, self- interest, and filial norms. The results from each area are synthesized and interpreted within the context of Taiwan’s modernization. The limitations of this study and the future directions of this study will also be given. Finally, the implications for public policy are discussed to highhght the contributions of this dissertation. Summary of Study Results In this study, there were two hypotheses being tested: 1) there are four possible types of intergenerational social support that exist between an older parent and any of their adult children, which are reciprocal exchange, upward transfers, downward transfers, and no exchange; and 2) there are three domains of motivation for intergenerational exchange that corresponds to a type of transfer. The upward transfers correspond to the following three models of motivation: altruism, power and bargaining, and filial norms. Income generation and corporate control models were expected to predict the downward transfers, while the mutual aid models of R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 138 motivation were hypothesized to predict the reciprocal exchange pattern of support. In the estrangement model, intergenerational transfers are expected to be non existent. With regard to the first hypothesis, there are indeed four possible intergenerational transfer patterns that exist in Taiwan; however, they are more detailed forms of upward transfer types. The first type is the older parent who receives emotionally or instrumentally from one or more adult children, known as the emotional and instrumental upward transfer class. The second type consists of the older parent who receives no support from their adult children, who belongs in the no upward transfer class. The next class consists of older adults who mainly receive fmancial and material support from their children, known as the economic upward transfer class. The final class is the full upward transfer class, which consists of older adults receiving across all categories of support from their children. Given Taiwan’s modernization and cultural values, one would expect that there would be a shift from traditional norms of care (full upward transfers) to more modemized patterns of support (reciprocal exchange). According to the LCA results, the emotional and instrumental upward transfer class had the highest probability of appearing in the Taiwanese population (34%), with the no upward transfer class at a close second (33%). The overall probabilities for the two classes were so close that the frequencies show that the no upward transfer class was actually the larger class that existed among Taiwanese families. This finding alone R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 139 shows that older adults and their children are grasping onto values similar to countries with a longer history of modernization, such as in the US and Europe. Based upon the four types of intergenerational transfer patterns that exist in the Taiwanese population, the seven models of motivations that predict these patterns are then tested. For the no upward transfer class, the following three types of models are tested: estrangement, corporate, and self-interest. For the rest of the three upward transfer classes, the following three models are tested as predictors: altruism, power and bargaining, mutual aid, and filial norms. Below is a discussion surrounding the outcomes of the four domains of motivation and their corresponding models of motivation. Need Domain The altruism model is supported by the data, as older adults who are suffering from poor health received economic transfers from their younger adult children. As hypothesized, those who are in poorer health are more likely to become receivers of some form of economic, emotional, or instrumental support, rather than no support. Interestingly, poor health does not significantly predict emotional and instrumental support, which is perhaps due to the implementation of the national health care system, which provides home health care, day care, and other long-term care services for the elderly. In 1999, 253 hospitals provided home health care to elderly who were discharged from hospitals but still required medical attention, or those who are chronically ill (GIO, 2001). The economic support is more important R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 140 than emotional and instrumental support because in order to pay for the necessary medical services, each elderly will need to pay approximately US$56 per visit to receive regular medical care about twice per month (GIO, 2001). In addition the Department of Health also provides out-patient services, day cares, nursing homes, and home care for cancer patients, and supportive services to family caregivers for chronically ill elderly (GIO, 2001). The provision of these services for elderly with poor health allows for relief of adult children from emotional and instrumental care, but not economic support. Consistent with other studies that found being older and female will be in need of more support (Hermalin, Ofstedal, & Lee, 1992; Kao & Stuifbergen, 1999; Lee, Parish, & Willis, 1994; Lee & Xiao, 1998; Shyu, Archbold, & Imle, 1998; Yamazaki, Weinstein, & Chang, 2000), this study did not find that being single, low socioeconomic status, living in rural areas and not receiving any income is significantly related to support. As stated in chapter 3, empirical evidence for the relationship between the parent’s health status and receiving financial support from their children has been mixed. In support of Lee and Xiao’s (1998) findings, but contrary to Hermalin, Ofstedal and Lee (1992), Lin and colleagues (2002), and Yang (1996), this study found that there is a strong relationship between health and economic support. However, contrary to what was expected though, being on welfare was not significant motivators in the altruism model for upward transfers. One explanation for this is offered by Lin and colleagues (1999) that the parents’ receipt R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 141 of support depends more on the children’s economic status, rather than their own. Furthermore, Biddlecom and colleagues (2001) conducted a study using previous waves of this data to show that elderly welfare programs provided by city and county-level governments have little effect on the economic support provided by the family to the older adult. Hence, altruistic giving from the younger generation in the Taiwanese population is motivated by health, rather than economic, needs of the older person. Exchange Domain The hypothesis for this domain is that older parents will receive transfers from their children in whom they give immediate care or have invested more in the past, such as those with higher education and those with potentially the highest socio-economic status. As mentioned in chapter 3, the mutual aid model emphasized reciprocity, suggesting that the extent to which adult children provide support for their parents is based upon the parents’ past investment in them (Gouldner, 1960; Thibaut & Kelley, 1959). For this analysis, the past and immediate investments are both tested in the same model under the exchange domain and were found to be inconsistent with the predictions that older adults would receive some type of support from their children for the services they provided. In fact, this model is not confirmed for immediate support of providing babysitting services and monetary aid and past investments in housing, studying abroad, and medical R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 142 treatment for a child. When compared to older adults who did not invest in their children, those who are giving support to the younger generation are less likely receivers of support. In a longitudinal analysis conducted by Lin and colleagues (2003), children who received a greater investment over ten years, such as tuition for education or fmancial gifts from parents, are more likely providers to their parents. Perhaps it is too early to see the returns from the investments made in the past three years for the adult children. According to Becker and Tomes (1976), the model of parental investment in their children’s human capital states that the amount of investment of parental time and expenditures for their child’s health and education increases the likelihood of receiving future returns from that child during adulthood. But based on this cross-sectional analysis, parents who adhere to higher levels of filial norms or do not offer immediate or precedent mutual aid are likely receivers of emotional and instrumental care. In terms of the sociodemographic characteristics, the old-old mother who is single, of Holo or Hakka descent, living in a rural area is a likely receiver of emotional and instrumental support from her son who is either single or who has had at least a college education, and to whom she has not offered past or immediate investments. Resources of Power This domain has two competing sub-hypotheses that older parents would receive transfers from their adult children based upon the economic power of R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 143 the older parent to promise future inheritance of property and assets. While adult children who abide by their parent’s economic requests and who are dependent upon their parents as the main wage earners are most likely receivers of downward transfers. The basis for the first sub-hypothesis is the power and bargaining model, which says that those who have either partially or not yet divided property to their children are likely receivers of support because according to Lee, Parish and Willis (1994), these parents promise future bequests and inheritance of property to their children in order to receive support from them. This first sub-hypothesis is confirmed for those who have divided some or none of their property is receiving both emotional and instrumental support or only economic aid from their adult children. However, in addition to property availability, those who are the main economic decision makers and main wage earners in the household are likely receivers of emotional and instrumental care from their adult children. This is in contrast to Lee, Parish, and Willis’ (1994) interpretation that older parents are losing power to extract obedience and resources from their children because they are faced with declining control over the family economy, such as their farms or businesses, as well as having skills that are quickly becoming obsolete with rapid economic development. Instead, the addition of these other two components of the resources of power domain—economic decision maker and main wage earner—show that older parents who hold both current and future bequests to their adult children, will R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 144 receive only the emotional and instrumental type of support because they are already providing economic support to their dependent adult children. According to Emerson (1972a), in the power-dependence balance, the struggle for older parents to be economically independent from, rather than dependent upon, their children will lead to maintenance of power to command support from them. The power and bargaining model is supported in this type of support for emotional and instrumental aid with the inclusion of all resources of power. Although these two other components of the domain were hypothesized to predict downward transfers, as part of the second sub-hypothesis, older adults who have control over economic decision making for the household are instead, receivers of all types of upward transfers. While those who are main wage earners with no property to divide, are more likely non-receivers. Contradictory to predictions, those who only have corporate control over the household are more likely receivers of all types of transfers. The corporate model emphasizes that the oldest male who has economic control over the household will give to the child most in need and who has attained the highest education or socioeconomic status (Lillard & Willis, 1997) because they are likely going to give back to them in the future when they are older. This cross-sectional finding suggests that the returns are more immediate as those who have corporate control over the household are likely receiving all types of care from a working daughter who is married. Although the original emphasis of the corporate model is on the paternal control over the welfare of the family members by maximizing their well-being and preservation through R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 145 intergenerational coresidence (Fricke et al, 1994; Thornton et al, 1994; Lee & Xiao, 1998), this study shows that the older females are more likely receivers of emotional, economic, and instrumental help when compared to older males. Under the self-interest model, older adults are assumed to provide to adult children who are wealthier and promising more to them in the future, as an insurance model (Lee & Xiao, 1998; Lillard & Willis, 1997). This model predicts no upward transfers for older parents who are the main wage earners in the household. Although they are receivers of emotional and upward transfers, they are not receivers of economic support. These are the older adults who have no available property to divide among their children in the future, which can be interpreted as a motivation for older adults to give to their children at the present time as insurance so that they can retrieve future support from their children when they are most in need. Ideology Domain The aim of this study is to test whether or not the social power of the older person determines their upward flow of support through the filial norms model. This model is confirmed among older single females, of Holo or Chinese ethnic descent, and who are living in rural areas. These older adults who have stronger feelings of normative values are more likely receivers of emotional and instrumental upward transfers. Furthermore, the characteristics of the adult children who are likely to give emotionally and instrumentally are those who are only daughters, at younger ages, with at least a college education, and who are also married. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 146 The ethnic differences in 1999 show that daughters of Holo or Chinese will be more likely to give both emotional and instrumental transfers to their older parent, when compared to the indigenous tribes. This is the only study that compares patterns of support between Chinese and indigenous tribes. Furthermore, it separates the Holo and Hakka groups from each other. Other studies using the same sample combine the Holo and Hakka groups together as the reference group and find that the Chinese are less likely to provide economic, emotional, or instrumental support between 1989 and 1999 (Biddlecom, Chayovan, & Ofstedal, 2002; Comman et al., 2001; Lin et al, 2002, 2003). However, between Holo and the indigenous group, there have been changes in the support patterns over the past decade, based upon a study by Lin and colleagues (2003), using the 1989 data, daughters of indigenous descent are more likely than daughters of Holo descent to provide instrumental and financial support. Ten years later, this study reveals that taking into account the emotional component and the fact that there are more coresidence between Holo generations than indigenous (37% vs. 23%, in respective order), shows a reversal in the patterns of daughters giving to their parents between these two ethnic groups. As mentioned before that prior studies have combined the Holo and Hakka ethnic groups together, which found that they have a higher likelihood of receiving support across all categories (Biddlecom, Chayovan, & Ofstedal, 2002; Comman et al, 2001; Lin et al, 2002,2003). When examining Holo and Hakka groups separately, it is the Holo, rather than the Hakka, ethnic group that R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 147 has the higher likelihood, when compared to the indigenous group, of providing emotional and instrumental care. In terms of the older adult’s economic resources, those who have a higher likelihood of obtaining emotional and instrumental upward transfers those who have either partially or not yet divided property, who is a main economic decision maker, and who earns and provides their own income for the household. This last economic predictor concerning income provision is an interesting finding for this filial norms model. Therefore, consistent with the second hypothesis, those with higher levels of normative expectations are more likely to be receivers of support. However, contrary to expectations that older adults are likely to receive economic aid from the adult child as the predominant support pattern, these older adults are receiving emotional and instrumental support instead because they are economically independent. The filial norms model shows that while taking into account the sociodemographic characteristics and economic resources of the older parents and their children, the older adult’s perceptions of social norms have a significant effect upon the pattern of support that the older adult receives. These perceptions for older people can change as explained by the theoretical framework proposed in this dissertation. As older adults experience the rapid pace and effects of modernization, which are guided by cultural values within a country, their normative values are also undergoing transformation that are parallel with other modemized countries. It has been assumed from the literature that there are more adult children who express filial R e p ro d u c e d with p erm issio n of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 148 piety in nontraditional ways by providing financially and materially, while living apart from their older parents (Biddlecom, Chayovan, & Ofstedal, 2002; Lin et al., 2002). However, among older adult’s who have maintained their own economic resources and traditional perceptions of filial piety, this study shows that they are more likely to become receivers of emotional and instrumental care from their children. Estrangement Model It was hypothesized that older Taiwanese adults who only have biological ties with their adult children will not engage in any transfers. However, this model is not supported in the data, as those who did give babysitting or fmancial help to their adult children were more likely, than those who do not give any support, to receive nothing in return from their adult children. The explanation for this pattern of downward giving will be further discussed in the self-interest model. What motivates intergenerational transfers? From this analysis, the domains of motivation are inferred for each type of social support. The resources of power and ideology of the older parent motivate emotional and instrumental support. There are no motivations behind older parents who do not receive any support, but instead they offer past and immediate support to their adult children and are economically independent. For economic support, older individuals perceive that they receive this type of support based upon R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 149 both health need and property ownership under resources of power. In terms of receiving emotional, economic, and instrumental support, older adults perceive the motivation for their children to provide for them based upon corporate control as the older adult is the major decision maker in the household. There is continuity of traditional norms in Taiwan, amidst the changes that occurring both on the macro- and micro-social levels. The presence of filial norms is still present as seen in the emotional and instrumental support received from older adults who have resources of economic power. Although many researchers have hypothesized that the skills and knowledge of older adults are becoming obsolete and they are losing power in society to command support from their adult children, unless they have property and assets available promised to their children in the future (Goode, 1963; Hermalin & Myers, 2002; Kinsella & Velkoff, 2001; Lee, Parish, & Willis, 1994; Lee & Xiao, 1998), this is only partially confirmed. It appears that older adults who are already providing economic support to the household as main wage earners at the time of the study hold onto the power to command support from their children in the form of emotional and instrumental help. In Confucian culture, filial piety is expressed by adult children under all circumstances by offering care to their older parents in coresidence, but it appears that filial norms is expressed towards a more modemized group of older adults who hold economic resources of power without proximity, similar to more modemized countries (Mangen & Westbrook, 1988). R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 150 Limitations and Future Directions o f the Study This study has several limitations. One weakness is that because of the potential complexity of the analysis, cross-sectional data was used from the 1999 wave. Hence, previous patterns of intergenerational support are not used in this analysis to account for differences across time. An additional inadequacy is that information regarding children’s provision of support is obtained from the parent. Parents may recall their interactions with some children better than those with other children. Another disadvantage of the survey is the inability to identify children matched to their spouses. This is especially important in a study on social support, since it is traditionally the daughter-in-law married to the oldest son who cares for her husband’s mother and father. Thus, if further information were provided regarding the division of support among the sons and daughters-in-laws, then this would add more details to this typology. A fourth constraint is the inabihty to identify which grandchild belongs to which child, preventing the researcher from understanding whether or not grandchildren are a competing responsibility that may reduce the flow of support from adult children to parents. An additional limitation is more detailed information regarding the children’s employment schedules, income, and social networks, which are possible influences upon their ability or inability to provide for their parents. R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth er reproduction prohibited w ithout perm ission. 151 Nevertheless, this study provides two important insights for aging research in Taiwan. First, the classification of social support types that are simultaneously given and received between older parents and their adult children through the use of LCA, with the full exchange type as most commonly found in the Taiwanese population. Second, based upon this typology, four models of motivations are found to exist in the population: 1) altruism model, which is based upon social and health needs of the older person; 2) power and bargaining model, which is based upon the older person’s economic decision making power and income from work; 3) self-interest model, which is based upon the economic resource of the older person as the primary income supporter for the household ; and 4) the filial norms model, which is based upon the older person’s perception of norms. Upon closer examination of the filial norms model, older adults who perceive higher levels of normative expectations are more likely to receive emotional and instrumental support from their children. In the near future, this study can be extended to include previous waves of the Taiwan data to perform longitudinal analyses within the context of causal modeling to enable the researcher to study individual changes over time. Collins and Flaherty (2002) use latent transition analysis, which is a variant of the latent class model that captures the "true" (latent) changes by analyzing stage- sequential developments in categorical characteristics. In addition, structural equation modeling can be used to examine the latent constructs for each model of R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 152 motivation. For example, comparing the economic, social, and health needs as latent constructs of the altruism model over each wave of data available as predictors for the three types of upward transfer models found through LCA. Implications for Family Theory The contribution of this analysis to intergenerational transfer theory for families is mainly seen through the resources of power and ideology domains. In all cultures, whether it is Asian or Western, these micro-social models of motivations for intergenerational support can be tested. Although the ideology domain may not be filial piety, in the case of non-Asian and non-Confucian countries it can take the form of familial norms, which Mangen and Westbrook (1988:188) defines as “standards of behavior that govern and mediate intrafamilial interactions and expectations of the family as a social system.” In the US, for example, American families are distinguished by norms of assistance in times of need, but not coresidence or within the same neighborhood (Kerchoff, 1966a, 1966b). As many scholars have noted that the conceptual bases for parent-child relations need to be enhanced, especially in the theoretical developments of norms (Bengtson et al., 1985; Hagestad, 1982). This analysis hope to add to the investigation surrounding norms and familism in Asia, as well as in non-Asian countries under which ideology can be used as lens to examine micro-social intergenerational exchanges that take place. R e p ro d u c e d with p erm issio n of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 153 The power of resources domain is also a contribution as it shows how different sub-domains of economic power are a major motivator for corresponding types of upward transfers. It appears to be a strong motivator when introduced into the equation, especially with large odds ratio for corporate control predicting full upward transfers. This may also add to theories related to economics and the family, which can help explain what types of transfers occur and why people give and receive from each other within the family. In terms of the macro-social level, the modified modernization framework helps to explain these micro-social models of intergenerational exchange in the context of rapid changes in demographic and economic advancements made within a country. For Taiwan, that is shifting towards being fully modemized and having more formal programs available, there is a larger group of older people who are not receiving any upward transfers, but instead, giving downward transfers to their adult children. This is very similar to countries, such as the United States and Sweden, where the value system leans more toward the self-expression pole, rather than just survival (Inglehart & Baker, 2000). Taiwan is becoming more like these “Westemized” countries having a more affluent older generation that gives downward transfers to their offspring (Chow, 2002). Older Taiwanese with more economic resources can still command filial piety in a nontraditional form from their adult children as Taiwan is experiencing the intersection of an increasingly capitalistic economy with changing cultural norms is predicted by Inglehart and Baker (2000). R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth er reproduction prohibited w ithout perm ission. 154 Implications for Social Policy in Taiwan In conclusion, the rapid demographic, economic and social changes over the last few decades, have shifted the traditional patterns of upward flows to a modemized pattern of interactive flows of support across generations with parents giving to, not just receiving from, their adult children in Taiwan. Part of the reason for this shift can be attributed to the economic prosperity experienced in Taiwan during the past decade, weakening the traditional family-based agricultural economy and the support for the elderly members, who no longer share in the risks of income loss or gains in profit from the service-based economy. With issues that come forth as a result of population aging, particularly declines in fertility and high rates of out migration by younger generations, the Taiwanese govemment must face the increasing likelihood that today’s parents will have only one unmarried child present during their older years who will take care of and be taken care of by the older parent, as shown in this study. Most importantly, the prevailing assumption in Asian countries that the oldest son will provide full support to their older parents is no longer the expected norm, in fact there are higher normative expectations and more intergenerational exchanges of instrumental and emotional care that takes place between older parents and their only daughters found in this study. The Taiwanese government has taken steps toward designing various social policies to enhance the well being of older adults. The universal health care system, which was implemented in 1995, has made health care accessible to all older R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 155 adults who choose to enroll in the program. However, this study shows that adult children are still involved in giving monetarily and materially to their older parents who in poorer health. As of 2002, a national elderly pension program was newly launched giving the older adult US$85 each month. To be eligible, they must meet both age (age 65 and over) and means-tested criteria (i.e., incomes less than $14,285). Although there are approximately 22% of the elderly who are eligible for this national program, enrollment has not been popular either because of the welfare stigma associated with pension program or govemment workers need to do more outreach about the program (GIO, 2003). Based upon this study, the older indigenous parents are least likely to receive emotional and instmmental support when compared to the Holo and Chinese, which may require more outreach efforts by the govemment in providing social services for these older adults living in traditional villages located in mountainous areas of Taiwan. Upon examining local levels of govemment welfare programs for the elderly, there are minimal effects upon the shift from private to public transfers, whereas the farmers pension program had a significant and positive effect of making this switch in 1996 (Biddlecom et al., 2002). Further analysis will need to be made about the implementation of this new pension system upon the elderly population and their intergenerational support patterns, in order to see if an expansion of the program to include all elderly is possible and necessary. As mentioned in the modified modemization framework and contrary to Cowgill’s (1972) modemization theory, the effects of modemization have given R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 156 younger and older generations more opportunities for economic and social independence. Based upon the findings from this dissertation that economic transfers from the older to younger generations still occur, provides evidence that older generations who are the main wage earners in their household still perceive their children as adhering to filial piety when they provide emotional and instmmental care. However, contrary to Confucian ideals, parents in old age are mainly providers in Taiwan, based upon the predictor of being the main wage earner in the household and as the no upward transfer pattern is the predominant group found in the dissertation. Thus, it appears that filial piety in the traditional sense is no longer the norm, but one that is modified where parents are the primary wage earners. Thus, the following questions arise: Is the type of support in Taiwan still called filial piety? Or is this similar to the westem model of intergenerational solidarity, as proposed by Bengtson and colleagues (1975)? Among the current generation of older adults in Taiwan, the notion of filial norms is still in existence. However, it is likely that future generations of older adults will be more like the intergenerational solidarity model, which will be an interesting factor to study in the intergenerational social support patterns for the future of Taiwanese families and other parts of Asia undergoing similar demographic, economic, and social trajectories. R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth er reproduction prohibited w ithout perm ission. 157 References Agree, E., Biddlecom, A.E., Chang, M.C., and Perez, A.E. (1998). Generalized exchange and intergenerational transfers in Taiwanese and Filipino families. Papers on Population, WP: 98-06. Hopkins Population Center, Johns Hopkins University. Agree, E., Biddlecom, A.E., Chang, M.C., and Perez, A.E. (2002). Transfers from older parents to children in Taiwanese and Filipino families. Journal o f Cross-Cultural Gerontology, 17,269-294. Aldenderfer, M.S., & Blashfiedl, R.K. (1984). Cluster Analysis. Beverly Hills, CA: Sage. Bartlett, H. and Wu, S. (2000). Ageing and Aged Care in Taiwan. In D.R. Phillips (Ed.), Ageing in the Asia-Pacific Region: Issues, Policies and Future Trends. London: Routledge. Beckett M. Goldman N. Weinstein M. Lin I.E., & Chuang YL. (2002). Social environment, life challenge, and health among the elderly in Taiwan. Social Science & Medicine, 55, 2, 191-209. Bengtson, V., Dowd, J., Smith, D., & Inkeles, A. (1975). Modernization, modernity, and Perceptions of Aging: A Cross-Cultural Study. Journal of Gerontology, 30, 6, 688-695. Biddlecom, A., Chayovan, N., & Ofstedal, M.B. (2002). Intergenerational Support and Transfers. In A. I. Hermalin (Ed.), The Well-Being o f the Elderly: A Four-Country Comparative Study (pp. 185-229). Ann Arbor, MI: The University of Michigan Press. Biddlecom, A., Hermalin, A., Ofstedal, M.B., Chang, M.C., & Chuang, Y.L. (2001). Tradeoffs between Public and Private Economic Support o f the Elderly: Results from a Natural Experiment in Taiwan. (Comparative Study of the Elderly in Asia Research Reports, Report No. 01-58). Population Studies Center at the Institute for Social Research, University of Michigan. Birkelund, G.E., Goodman, L.A., & Rose, D. (1986). The latent structure of job characteristics of men and women. American Journal o f Sociology, 102, 80- 113. R e p ro d u c e d with p erm issio n of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 158 Bordy, C.J. (1986). Things are rarely black and white: Admitting gray into the Converse model of attitude stability. American Journal o f Sociology, 92, 657-677. Campbell, W. (1903). Formosa under the Dutch. London: Kegan Paul, Trubner & Co. Castle, D.J., Sham, P. C., Wessley, S., & Murray, R.M. (1994). The subtyping of schizophrenia in men and women: A latent class analysis. Psychological Medicine, 24, 41-51. Chan, H.S. (1992). Ageing in Taiwan. In D.R. Phillips (Ed.), Ageing in East and South-east Asia. London: Edward Arnold. Chang, T.P. (1994). Family changes and the elderly in Asia. In Department of Economic and Social Information and Policy Analysis (Ed.), The Ageing o f Asian Populations. (Proceedings of the United Nations Round Table on the Ageing of Asian Populations, No. 125). New York: United Nations Publication. Chao, P. (1983). Chinese kinship. Boston: Kegan Paul International. Chen, X. & Silverstein, M. (2000). Intergenerational social support and the psychological well-being of older parents in China. Research on Aging, 22, 1, 43-65. Clogg, C.C., & Goodman, L.A. (1984). Latent structure analysis of a set of multidimensional contingency tables. Journal o f the American Statistical Association, 79, 762-771. Clogg, C.C., & Goodman, L.A. (1985). Simultaneous latent structure analysis in several groups. In N.B. Tuma (Ed.) Sociological methodology (pp. 132-193). San Francisco: Josey-Bass. Clogg, C.C., & Goodman, L.A. (1986). On scaling models applied to data from several groups. Psychometrika, 5 1 ,123-135. Comman, J.C., Goldman, N., Weinstein, M., & Lin, H.S. (2001). Perceptions o f elderly Taiwanese about the availability o f social support (Office of Population Research Working Paper Series, No. 2001-05). Princeton, NJ: Princeton University. R e p ro d u c e d with p erm issio n of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 159 Davidson, J.W. (1903). The island o f Formosa, past and present. London: Macmillan & Co. Directorate General of Budget Accounting and Statistics. (1999). Monthly Bulletin o f Statistics o f the Republic o f China. Taipei, Taiwan: Executive Yuan. Domingo, C. & Asis, M. (1995). Living arrangements and the flow of support between generations in the Philippines. Special Issue: Focus Group Research on the Living Arrangements of the Elderly in Asia. Journal o f Cross Cultural Gerontology, 1 0 ,21-51. Easterlin, R.A., Schaeffer, C.M., & Macunovich, D.J. (1993). Will the baby boomers be less well off than their parents? Income, wealth, and family circumstances over the life cycle in the United States. Population and Development Review, 19, 3, 497-522. Fergusson, D.M., Horwood, J.L., & Lynskey, M.T. (1994). The comorbidities of adolescent problem behaviors: A latent class model. Journal o f Abnormal Child Psychology, 22, 339-354 Freedman, R., Thornton, A., & Yang, L.S. (1994). Determinants of co-residence in extended households. In A. Thornton & H.S. Lin (Eds.), Social change and the family in Taiwan (pp.335-358). Chicago: University of Chicago Press. Fricke, T., Liu, P.K.C., Thornton, A., Freedman, D., & Yang, L.S. (1994). The changing organization of individual activities. In A. Thornton & H.S. Lin (Eds.), Social change and the family in Taiwan (pp. 116-147). Chicago: University of Chicago Press. Govemment Information Office. (2002). The Republic o f China Yearbook: Taiwan 2002. Taipei, Taiwan: Govemment Information Office. 2003. Govemment Information Office. (2003, August). Taiwan Yearbook 2003. Retrieved January I, 2004, from httpJ/vm w .gio.gov.tw/taiwaii-website/5- gp/vearbook/ Hermalin, A.I. (2002). Making the choices: Policies and research for the coming years. In A.I. Hermalin (Ed.), The well-being o f the elderly: A four-country comparative study (pp. 143-184). Ann Arbor, MI: The University of Michigan Press. Inglehart, R. & Baker, W.E. (2000). Modemization, cultural change, and the persistence of traditional values. American Sociological Review, 65, 19-51. R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 1 6 0 Kerr, G.H. (1974). Formosa, licensed revolution and the Home Rule Movement, 1895-1995. Honolulu: The University Press of Hawaii, Honolulu. 1895-1996. Kinsella, K. (2000). Demographic dimensions of ageing in East and Southeast Asia. In D. R. Phillips, Ageing in the Asia-Pacific Region: Issues, policies and fiuture trends. London: Routledge. Knodel, J. & Ofstedal, M.B. (2002). Patterns and Determinants of Living Arrangements. In A. I. Hermalin (Ld.), The well-being o f the elderly: A four-country comparative study (pp. 143-184). Ann Arbor, M I: The University of Michigan Press. Le, C. T. (1998). Applied Categorical Data Analysis^ New York: John Wiley & Sons, Inc. Lee, Y., Parish, W.L., & Willis, R. (1994). Sons, daughters, and intergenerational support in Taiwan. American Journal o f Sociology, 99, 4, 1010-1041. Lee, Y. & Xiao, Z. (1998). Children’s support for elderly parents in urban and rural China: Results from a national survey. Journal o f Cross-Cultural Gerontology, 13, 39-62. Li, R. (1994). Aging Trends—Taiwan. Cross-Cultural Gerontology, 9, 389-402. Lillard, L. & Willis, R. (1997). Motives for intergenerational transfers: Evidence from Malaysia. Demography, 34, 1, 115-134. Lin, H.S., Lee, M.L., & Thornton, A. (1994). Trends in the timing and prevalence of marriage. In A. Thornton & H.S. Lin (Eds.), Social Change and the Family in Taiwan (pp. 202-224). Chicago: University of Chicago Press. Lin, I.F., Goldman, N., Weinstein, M., Lin, Y.H., (2002). Stability and change in patterns o f intergenerational transfers in Taiwan. Paper presented at the International Meeting on Age Structure Transitions and Policy Dynamics: The Allocation of Public and Private Resources across Generations, Taipei, Taiwan. Lin, I.E., Goldman, N., Weinstein, M., Lin, Y.H., Gorrindo, & T. Seeman, T. (2003). Gender differences in adult children’s support of their parents in Taiwan. Journal o f Marriage and Family, 65, 184-200. Mason, A., Lee, S., & Russo, G. (2000). Population momentum and population aging in Asia and Near-East countries. Honolulu, HI: East-West Center. R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 161 Mitchell, C. & Plunkett, M. (2000). The latent structure of substance use among American Indian adolescents; An example using categorical variables. American Journal o f Community Psychology, 28, 1,105-125. Molm, L.D., Takahashi, N., & Peterson, G. (2000). Risk and trust in social exchange: An experimental test of a classical proposition. American Journal o f Sociology, 105, 5, 1396-1427. Ofstedal, M.B, Chan, A., Chayovan, N., Chuang, Y.L., Perez, A. Mehta, K., & Hermalin, (2002). Policies and programs in place and under development. In A. I. Hermalin (Ed.), The well-being o f the elderly: A four-country comparative study (pp. 65-100). Ann Arbor, MI: The University of Michigan Press. Olds, Kelly. The economic history of Taiwan. In R. Whaples (Ed.), EH.Net Encyclopedia, Retrieved March 24 2003 from http://www.eh.net/encvclopedia/contents/olds.taiwan.economic.historv.php Phillips, D.R. (2000). Ageing in the Asia-Pacific Region: Issues, policies andfuture trends. London: Routledge. Randel, J., German, T. & Ewing, D. (Eds.) (1999). The ageing and development report: Poverty, independence and the world’ s older people. HelpAge International, London: Earthscan. Shi, L . (1993). Health-Care in China: A rural-urban comparison after the socio economic reforms. Bulleting o f the World Health Organization, 39, 324-331. Shigemasu, K., & Sugiyama, N. (1994). Latent class analysis of choice behavior. Behaviormetrika. 21. 49-60. Silverstein, M., Parrott, T., & Bengtson, V. (1995). Factors that predispose middle- aged sons and daughters to provide social support to older parents. Journal o f Marriage and the Family, 57, 465-475. SPPS, Inc. (1993). SPSS (Version 11). Chicago, IL: SPSS, Inc.99. Tamney, J.B. & Chiang, L.H. (2002). Modernization, globalization, and Confucianism in Chinese societies. Westport, CT: Praeger. R e p ro d u c e d with p erm issio n of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 162 Weinstein, M., Sun, T.H., Chang, M.C., and Freedman, R. (1994). Co-residence and other ties linking couples and their parents. In A. Thornton & H.S. Lin (Eds.), Social Change and the Family in Taiwan (pp. 305-334). Chicago: University of Chicago Press. Wu, S. & Chu, C. (1996). Public attitudes toward long-term care arrangement for the elderly in Taiwan. Australian Journal on Ageing, 15, 2, 62-68. R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 163 Appendix A Odds Ratios from Multinomial Logistic Regression for Types of Support Transfer Comparisons (N=9477) Table A1. Odds ratios from multinomial logistic regression of being In emotional and Instrumental upward transfer class vs. no upward Variables Model 1 Model 2 Model 3 Model 4 Soclodem oaraohic Characteristics o f Resoondents Age group Age 65-74 1.24 1.28 1.16 1.14 Age 75-84 1.45* 1.45* 1.38 1.36 Age 85+ (reference) Gender Female 1.25*' 1.27*** 1.44*** 1.44*** Male (reference) Marital Status Married or living with a 0.56*** 0.58*** 0.57*** partner Single, divorced, widowed (reference) Ethnic Holo 0.55*** 1.54* 1.55* 1.64* 1.64* Hakka 1.43 1.43 1.50 1.51 Chinese 1.79** 1.84** 1.91** 1.89** Indigenous (reference) Place of Residence Rural 1.38*** 1.39*** 1.31*** 1.33*** Urban (reference) Literate 1.07 1.09 1.10 1.09 Illiterate (reference) Median quartile income 0.99 0.98 0.90 0.90 Upper quartile income 0.95 0.95 0.86 0.84 Lower quartile income (reference) Has pension 1.13 1.15 1.09 1.16 No pension (reference) Sociodemographic Characteristics of Children Age group 12 through 18 (reference) 19 through 34 0.42 0.48 0.46*** 0.46*** 35 through 54 0.60*** 0.62*** 0.63** 0.63** 55 and over 0.56*** 0.55*** 0.57** 0.58** Birth Order Oldest child 0.97 0.86 0.86 0.86 Youngest child 0.76** 0.68** 0.70** 0.70** In Between 0.51*** 0.46*** 0.47*** 0.47*** Only child (reference) R e p ro d u c e d with p erm issio n of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 164 Table A1. Odds ratios from multinomial logistic regression of being in emotional and instrumental upward transfer class vs. no upward transfer class (comparison group) (N=9477) (continued) Variables Model 1 Model 2 Model 3 Model. Gender 1.19* 1.19* 1.19* Female 1.24* Male (reference) Grandchild 1.17* At least one 1.12 1.19 1.19 None (reference) Place of Residence Live next door to 0.49 0.45 0.44 0.44 parent Live in the same 0.42 0.40 0.39 0.40 neighborhood Live elsewhere in 0.43 0.42 0.40 0.40 Taiwan Lives overseas 0.35 0.36 0.35 0.34 Coresidence (reference) Education Elementary and less (reference) High school 1.09 1.13 1.12 1.13 College 1.20* 1.21** 1.17* 1.17* Work Full time 0.91 0.94 0.95 0.95 Part time 2.37 2.20 2.08 2.07 Unemployed (reference) Marital Status 1.20* Married or living with a 1.38**’ 1.43*** 1.44*** partner Single, divorced, widowed (reference) Frequency of face-to-face contact Daily, coresidence (reference) Daily, noncoresidence 1.75 1.77 1.64 1.66 Weekly, noncoresidence 1.09 1.04 0.99 1.01 Monthly,noncoresidence 0.84 0.79 0.76 0.77 Yearly, noncoresidence 0.73 0.68 0.65 0.66 Frequency of phone contact Daily, coresidence (reference) Daily, noncoresidence 1.10 1.12 1.13 1.11 Weekly,noncoresidence 0.85 0.86 0.86 0.84 Monthly,noncoresidence 0.74 0.74 0.73 0.71 Yearly, noncoresidence 0.74 0.77 0.79 0.78 Need: Altruism Health Need: Self-rated health Good health 0.99 1.00 0.97 0.97 Average health 0.99 1.00 0.98 0.98 Poor health(reference) Economic Need: Welfare status Enrolled in elderly welfare 1.02 0.99 1.01 1.09 program Not enrolled in elderly welfare program (reference) R e p ro d u c e d with p erm issio n of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 165 Table A 1. Odds ratios from multinomial logistic regression of being in emotional and instrumental upward transfer class vs. no upward transfer class (comparison group) (N=9477) (continued) Variables Model 1 Model 2 Model 3 Model 4 Exchanges: Mutual Aid Immediate Babysit(No=reference) 0.53*** 0.52*** 0.53*** Money (No=reference) 0.58** 0.56** 0.56*' Precedent House, education abroad, or 0.45*** 0.42*** 0.42*** medical treatment No past investments (reference) Resources o f Power. Power and Bargaining Property ownership No property to divide (reference) All property divided 1.20 Partiaiiy divided property 1.37 ^ Not yet divided property 1.24 * Corporate Control Main economic decision maker 1.16* (No=reference) Income Generator Respondent is main wage earner for household 1.36*** Respondent is no main wage earner for household (reference) Ideology: Filial Norms Levels of norms Wald (df) AX^ (Adf) % Variance Explained (PseudoR^) % A Log Likelihood -2 Log Likelihood A -2 Log Likelihood *** p<J01, **p<.01, ’ 1358.68 ( 120)*** 8.19% -9521.83 19043.66 37902.43 (129)**' 36543.75 (9)" 8.84% 0.65% -9454078 18931.74 111.92 30520.44 (144)*** 7381.99 (15)*'* 9.54% 0.70% -9382.47 18764.94 166.80 1.21 1.38*' 1.24*' 1.15* 1.35 1,02 43867.96 (147)*" 13347.52 (3)*** 9.60% 0.06% -9376.34 18752.68 12.26 p<.05 R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 166 Table A2. Odds ratios from multinomial logistic regression of being in economic upward transfer vs. no upward transfer (comparison group) (N=9477) Variables Model 1 Model 2 Model 3 Model 4 Sociodemoaraohic Characteristics of Resoondents Age group Age 65-74 0.63* 0.65 0.66 0.66 Age 75-84 0.91 0.92 0.92 0.92 Age 85+ (reference) Gender Female 1.27* 1.25* 1.19 1.19 Male (reference) Marital Status Married or living with a 1.00 0.99 0.97 0.97 partner Single, divorced, widowed (reference) Ethnic Holo 1.30 1.33 1.26 1.27 Hakka 1.26 1.29 1.22 1.22 Chinese 1.32 1.38 1.34 1.35 Indigenous (reference) Place of Residence Rural 1.19 1.17 1.13 1.13 Urban (reference) Economic resources Literate 0.92 0.93 0.92 0.91 Illiterate (reference) Median quartile income 1.17 1.17 1.18 1.18 Upper quartile income 0.96 0.97 1.04 1.04 Lower quartile income (reference) Has pension 0.74 0.74 0.79 0.79 No pension (reference) Sociodemographic Characteristics of Children Age group 12 through 18 (reference) 19 through 34 0.48** 0.49** 0.55** 0.55** 35 through 54 0.62 0.62* 0.64* 0.64* 55 and over 0.83 0.81 0.83 0.83 Birth Order Oldest child 1.07 1.08 1.08 1.07 Youngest child 1.05 1.05 1.04 1.04 In Between 1.15 1.14 1.12 1.12 Only child (reference) Gender Female 1.08 1.09 1.07 1.07 Male (reference) Grandchild At least one 1.22 1.22 1.10 1.10 None (reference) Place of Residence Live next door to parent 1.91 2.12 1.87 1.87 R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 167 Table A2. Odds ratios from multinomial logistic regression of being in upward transfer vs. no upward transfer (comparison group) (N=9477) Variables Model 1 Model 2 Model 3 Live in the same 1.50 1.73 1.52 neighborhood Live elsewhere in 1.72 Taiwan Lives overseas 1.07 Coresidence (reference) Education Elementary and less (reference) High school 0.91 College 0.91 Work Full time 1.21 Part time 4.18 Unemployed (reference) Marital Status Married or living with a partner 1.13 Single, divorced, widowed (reference) Frequency of face-to-face contact Daily, coresidence (reference) Daily, noncoresidence 0.99 Weekly,noncoresidence 1.00 Monthly,noncoresidence 0.79 Yearly, noncoresidence 0.61 Frequency of phone contact Daily, coresidence (reference) Daily, noncoresidence 1.01 Weekly,noncoresidence 0.91 Monthly,noncoresidence 1.05 Yearly, noncoresidence 0.85 Models o f Motivation Need: Altruism Health Need: Self-rated health Good health 0.65** Average health 0.80 Poor health(reference) Economic Need: Welfare status Enrolled in elderly welfare 1.04 program Not enrolled in elderly welfare program (reference) Exchange: Mutual Aid Immediate Babysit(No=reference) Money (No=reference) Precedent House, education abroad, or 0.37*** 0.35* medical treatment No past investments (reference) economic (continued) Model 4 1.52 1.99 1.78 1.77 1.29 1.22 1.21 0.91 0.93 0.93 0.92 0.93 0.93 1.22* 1.22* 1.22* 4.16 4.15 4.17 1.13 1.14 1.14 0.88 1.04 1,05 0.90 1.05 1.05 0.71 0.83 0.84 0.56 0.65 0.65 0.98 0.99 0.99 0.88 0.91 0.90 1.00 1.01 1.01 0.83 0.84 0.84 0.66** 0.67** 0.67 0.81 0.81 0.81 1.02 1.01 1.01 0.54*** 0.53*** 0.53* 0.45** 0.44** 0.44* 0.35 R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 168 Table A2. Odds ratios from multinomial logistic regression upward transfer vs. no upward transfer (comparison group) Variables Model 1 Model 2 Resources: Power and Bargaining Property ownership No property to divide (reference) All property divided Partially divided property Not yet divided property Corporate Control Main economic decision maker (No=reference) Income Generator Respondent is main wage earner for household Respondent is no main wage eamer for household (reference) Ideology: Fiiial Norms Levels of norms Wald (df) AX^ (Adf) % Variance Explained (Pseudo R ) % A Log Likelihood -2 Log Likelihood A -2 Log Likelihood * * * p<.001, **p<.01, *p<.05 of being In economic (N=9477) (continued) Model 3 Model 4 1358.68 ( 120)*'* 8.19% -9521.83 19043.66 37902.43 (129)*** 36543.75 (9)** 8.84% 0.65% -94540.78 18931.74 111.92 1.25 1.53* 1.20 1.28 0.63 30520.44 (144)*** 7381.99 (15)*** 9.54% 0.70% -9382.47 18764.94 166.80 1.25 1.53* 1.20 1.28 0.63 43867.96 (147)*** 13347.52 (3)*** 9.60% 0.06% -9376.34 18752.68 12.26 R e p ro d u c e d with p erm issio n of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 169 Table A3. Odds ratios from multinomial logistic regression of being in full Variables Model 1 Model 2 rz-A i:............... Model 3 Model 4 Sociodem oaraohic Characteristics o f Resoondents Age group Age 65-74 0.48 0.48 0.39* 0.38* Age 75-84 0.62 0.62 0.54 0.53 Age 85+ (reference) Gender Female 1.36 1.39 1.76* 1.76* Male (reference) Marital Status Married or living with a 0.45* 0.46*** 0.49** 0.48** partner Single, divorced, widowed (reference) Ethnic Holo 1.57 1.48 1.45 1.45 Hakka 1.05 0.97 0.96 0.97 Chinese 2.00 1.92 1.72 1.72 Indigenous (reference) Place of Residence Rural 2.07** 2.08** 1.98** 2.01** Urban (reference) Economic resources Literate 0.94 0.96 0.94 0.93 Illiterate (reference) Median quartile income 2.08** 2.07** 1.86* 1.84* Upper quartile income 1.46 1.51 1.39 1.37 Lower quartile income (reference) Has pension 0.80 0.81 0.79 0.79 No pension (reference) Sociodem ographic Characteristics o f Chiidren Age group 12 through 18 (reference) 19 through 34 0.28 0.34 0.34 0.34 35 through 54 0.59 0.61 0.66 0.65 55 and over 0.67 0.66 0.75 0.75 Birth Order Oldest child 1.27 1.09 1.05 1.05 Youngest child 0.84 0.73 0.75 0.75 In Between 0.80 0.69 0.68 0.69 Only child (reference) Gender Female 1.96** 1.92** 1.87** 1.88** Male (reference) Grandchild At least one 3.04 2.90 3.00 3.00 None (reference) Place of Residence Live next door to parent 0.77 0.70 0.71 0.71 R e p ro d u c e d with p erm issio n of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 170 Table A3. Odds ratios from multinomial logistic regression of being in full upward transfers vs. no upward transfer (comparison group) (N=9,477) (continued) Variables Model 1 Model 2 Model 3 Model 4 Live in the same 0.65 0.61 0.60 0.61 neighborhood Live elsewhere in Taiwan 1.17 1.13 1.05 1.06 Lives overseas 0.64 0.65 0.59 0.59 Coresidence (reference) Education Elementary and less (reference) High school 1.03 1.07 1.05 1.06 College 1.09 1.10 0.93 0.93 Work Full time 3.20*" 3.29*" 3.30"* 3.31' Part time 61.38" 51.55**" 51.87" 51.89* Unemployed (reference) Marital Status 1.96* 1.96* Married or living with a 1.59 1.82 partner Single, divorced, widowed (reference) Frequency of face-to-face contact Daily, coresidence (reference) 8.41* 8.28* Daily, noncoresidence 6.87 6.98 Weekly,noncoresidence 4.09 3.82 3.29 3.31 Monthly,noncoresidence 1.69 1.57 1.32 1.33 Yearly, noncoresidence 0.89 0.80 0.65 0.65 Frequency of phone contact Daily, coresidence (reference) Daily, noncoresidence 1.04 1.07 1.14 1.12 Weekly,noncoresidence 0.61 0.62 0.66 0.65 Monthly,noncoresidence 0.44 0.45 0.49 0.48 Yearly, noncoresidence 0.37 0.40 0.44 0.43 Models o f Motivation N ee d : Altruism Health Need: Self-rated health Good health 0.63 0.64 0.61 0.61 Average health 1.04 1.05 0.99 0.99 Poor health(reference) Economic Need: Welfare status Enrolled in elderly 1.38 1.34 1.40 1.39 welfare program Not enrolled in elderly welfare program (reference) Exchange: M utual A id Immediate Babysit(No=reference) 0.45 0.41 0.42 Money (No=reference) 0.86 0.84 0.83 Precedent Mutual Aid House, education abroad, or 0.20*** 0.44*** 0.35' medical treatment No past investments (reference) R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth er reproduction prohibited w ithout perm ission. 171 Table A3. Odds ratios from multinomial logistic regression of being In full upward transfers vs. no upward transfer (comparison group) (N=9,477) (continued) Resources Property ownership No property to divide (reference) Variables Model 1 Model 2 Model 3 Model 4 All property divided 0.88 0.89 Partially divided property 1.40 1.40 Not yet divided property 1.20 1.29 Corporate Control Main economic decision m aker 2 .2 6 * 2.24" (No=reference) Income Generator Respondent is main wage earner for household 0.92 0.92 Respondent is no main wage earner for household (reference) Ideoloav: Filial Norms Levels of norms 1.02 Wald 1358.68 37902.43 30520.44 43867.96 (df) (120)"* (129)”* (144)”* (147)”* AX^ 36543.75 7381.99 13347.52 (Adf) (9)" (15)*” (3)”* % Variance Explained (Pseudo R ) 8.19% 8.84% 9.54% 9.60% % A 0.65% 0.70% 0.06% Log Likelihood -9521.83 -94540.78 -9382.47 -9376.34 -2 Log Likelihood 19043.66 18931.74 18764.94 18752.68 A -2 Log Likelihood 111.92 166.80 12.26 ■ p<.001, **p<.01, *p<.05 R e p ro d u c e d with p erm issio n of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission.

Asset Metadata

Creator Yang, Frances M. (author)

Core Title Filial expectations and social exchange patterns among older Taiwanese parents and their adult children

Contributor Digitized by ProQuest (provenance)

School Graduate School

Degree Doctor of Philosophy

Degree Program Gerontology

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

Tag Gerontology,OAI-PMH Harvest,sociology, individual and family studies,sociology, theory and methods

Language English

Advisor Silverstein, Merril (committee chair), Bengston, Vern (committee member), Biblarz, Timothy (committee member), Crimmins, Eileen M. (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-532845

Unique identifier UC11340252

Identifier 3140578.pdf (filename),usctheses-c16-532845 (legacy record id)

Legacy Identifier 3140578.pdf

Dmrecord 532845

Document Type Dissertation

Rights Yang, Frances M.

Type texts

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

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA