xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001910878
007 cr mnu|||uuuuu
008 071002s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001717
Trede, Teri A.
An investigation of the relationship between religiosity and subjective well-being in older adults :
b the mediating role of optimism
h [electronic resource] /
by Teri A. Trede.
[Tampa, Fla] :
University of South Florida,
ABSTRACT: The proportion of adults aged 65 and older in the United States has beenincreasing steadily for years. Although most would agree that this increased longevity is are markable achievement, there is growing concern regarding the resources that will be necessary to provide care and services for this rapidly expanding segment of the population. From a public health perspective, simply increasing longevity is no longer the ultimate outcome sought. A more pertinent goal is increasing the quality of life years, or promoting resilience and successful aging. Strategies that assist adults in delaying the onset of disability or reducing the severity of disabling conditions may serve a vital role in the promotion of resilience and successful aging in older adults. The purpose of this study was to explore the association between religiosity and subjective well-being,commonly identified as a marker of resilience in older adults. In addition, the study sought to determine if that association is mediated by optimism. For this study, subjective well-being was operationalized with measures of depressive symptomatology and life satisfaction. Secondary analyses were conducted on longitudinal data collected for theFlorida Retirement Study. Results indicated that religiosity was not significantly associated with future depressive symptomatology, but was significantly associated with future life satisfaction. Dispositional optimism did not mediate the relationships between religiosity and subjective well-being variables.
Dissertation (Ph.D.)--University of South Florida, 2006.
Includes bibliographical references.
Text (Electronic dissertation) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 243 pages.
Adviser: Jeannine Coreil, Ph.D.
x Public Health
t USF Electronic Theses and Dissertations.
4 0 856
An Investigation of the Relationship Between Religiosity and Subjective Well-Being in Older Adults: The Mediating Role of Optimism by Teri A. Trede A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Community and Family Health College of Public Health University of South Florida Co Major Professor: Melinda S. Forthofer, Ph.D. Co Major Professor: Jeannine Coreil, Ph.D. Robert McDermott, Ph.D. Cathy McEvoy, Ph.D. Brent Small, Ph.D. Date of Approval: June 29, 2006 Keywords: Successful aging, resilience, lif e satisfaction, depressive symptomatology, retirement community Copyright 2006, Teri A. Trede
Dedication This dissertation is dedicated to the ma ny Â‘olderÂ’ adults that inspired me to undertake this odyssey Â– Rose, Evelyn, Hazel, Jack, Betty, Henrietta, Clayton, Hilda, Audrey, Floree, Dottie, Ken, Martha, Helen, Joyce, Lois, Betty, Dorothy and so many more. Though many of you have passed from this life, your reverence for life, courage, wisdom, compassion and selfless dedication to others will inspire me for a lifetime. And to my father, IÂ’m so sorry that your journey ended before mine Â– I so wish that you were here to celebrate this achievement with me.
Acknowledgements I would like to acknowledge and thank the individuals who have supported me throughout this process: Dr. Melinda Forthofer, whose patience, wisdom, guidance and endless support were invaluable. I can think of no finer mentor or role model. I am frequently in awe of your talent, your boundless energy, the enthus iasm with which you embrace change, and your adventurous spirit. And I am et ernally grateful for your friendship. Dr. Jeannine Coreil, who, as the first in structor I encountered at COPH, inspired me to pursue a degree in public health. I am very grateful for your guidance, support and encouragement. Dr. Robert McDermott, Dr. Cathy McEvoy, and Dr. Brent Small, IÂ’m grateful for your guidance and support. Your positive energy and dedication to your students and to your profession I admire greatly. And the friends who have shared this j ourney, whether you meant to or not, thank you for your support and friendship, Sandy & Bob Thursby, Dr. Caroline Mae McKay, Dr. Jodi Nearns, Dr. Lauren Bailey-Zapata, Becky Bullock, and Ja nice & Mark Peterson. And finally, to my three little boyfri ends, Paul, Daniel, & Benjamin, you have inspired me to search for knowledge, so that I might answer your questions, and challenged me to complete this work, so that I might graduate before you!
i Table of Contents List of Tables................................................................................................................. ...vii List of Figures................................................................................................................ .....x Abstract....................................................................................................................... .......xi Chapter One.................................................................................................................... ....1 Introduction.............................................................................................................1 Statement of the Problem........................................................................................1 The Graying of America.............................................................................2 Longevity vs. Quality of Life......................................................................5 Need for the Study..................................................................................................6 Implications for Public Health................................................................................7 The State of Gerontological Research on Aging....................................................9 Purpose of Study...................................................................................................11 Conceptual Model.................................................................................................11 Research Questions...............................................................................................12 Overview of Study Methods.................................................................................14 Research Hypotheses............................................................................................15 Delimitations.........................................................................................................15 Limitations............................................................................................................16 Definitions.............................................................................................................19 Chapter Two.................................................................................................................... ..22
ii Background...........................................................................................................22 Successful Aging..................................................................................................23 Early Theories...........................................................................................23 Expansion of Successful Aging Theories.................................................25 Current Views of Successful Aging..........................................................26 Gains and Losses in Aging.......................................................................27 Successful Adaptation...............................................................................28 Alternative Models....................................................................................30 Research Gaps...........................................................................................31 Resilience..............................................................................................................32 Defining Resilience...................................................................................32 Initial Resilience Research........................................................................33 Resilience and Adults...............................................................................33 Protective Factors Related to Resilience...................................................36 Resilient Adults.........................................................................................37 Reserve Resources and Resilience Outcomes...........................................39 Purpose of Study.......................................................................................39 Dependent Variables Â– Subjective Well-being.....................................................40 Identifying and Predicting Subjective Well-being....................................40 Subjective Well-being...............................................................................40 Life Satisfaction........................................................................................41 Depressive Symptomatology....................................................................42 Independent Variables..........................................................................................43
iii Resilience and Proactive Behaviors..........................................................43 Religiosity.............................................................................................................45 Prevalence of Religious Be havior and Affiliation....................................45 Religion and Well-being...........................................................................46 Age Effects on Religious Involvement.....................................................48 Possible Mechanisms of ReligionÂ’s Influence..........................................49 Negative Effects of Religion on Health and Well-being..........................51 Causal Considerations and Alternative Explanations...............................52 Measuring Religious Involvement............................................................53 Multidimensionality of Religious Involvement........................................54 Mediating Variable...............................................................................................55 Dispositional Optimism............................................................................55 Optimism as an Explanatory Factor in Resilience Outcomes...................57 Control Variables..................................................................................................58 Covariates.............................................................................................................59 Health Self Appraisal................................................................................59 Positive and Negative Affect....................................................................60 Functional Health / Physical Impairment.................................................62 Research Questions...............................................................................................63 Chapter Three.................................................................................................................. ..65 Methods.................................................................................................................65 Purpose of the Study.................................................................................65 Conceptual Framework for the Study.......................................................65
iv Research Hypotheses................................................................................66 Sampling and Data Collection Procedures...........................................................67 Data Source...............................................................................................67 Florida Retirement Study Sample.............................................................68 Participant Interviews...............................................................................70 The Current Study.....................................................................................71 Missing Data.............................................................................................72 Measures...............................................................................................................72 Dependent Variables Â– Subjective Well-being.........................................72 Life Satisfaction............................................................................72 Depressive Symptomatology........................................................75 Independent Variable................................................................................79 Religious Involvement..................................................................79 Psychometric Evaluation of Instruments......................................82 Mediators..................................................................................................85 Optimism.......................................................................................85 Confounders or Covariates.......................................................................88 Physical Health & Functional Ability...........................................89 Health Self-Appraisal....................................................................91 Positive and Negative Affect........................................................91 Sociodemographic Variables........................................................93 Data Analysis............................................................................................95 Univariate Analyses......................................................................96
v Bivariate Analyses........................................................................97 Analyses to Explore Possible Confounding by Health and Function Variables.......................................................................................98 Multivariate Analysis..............................................................................100 Modeling Issues..........................................................................101 Mediation Effects........................................................................102 Overview of Regression Model Building Process......................104 Design Limitations......................................................................106 Chapter Four................................................................................................................... 107 Study Sample......................................................................................................107 Characteristics of the Sample Population at Baseline.............................107 Physical Health and Function.................................................................114 Psychological Well-being.......................................................................114 Religiosity...............................................................................................115 Subjects Excluded Due to Missing Data at Baseline..........................................116 Exploring the Implications of Attrition...............................................................120 Comparing the Study Sample to the Elderly Population of Florida...................122 Bivariate Analyses..................................................................................126 Relationships between the Control Variables, Covariates, Optimism, Time 10 Life Satisfaction and Time 10 Depressive Symptomatology............131 Multivariate Analysis..........................................................................................138 Overview.................................................................................................138 Regression Diagnostics...........................................................................140
vi Building the Regression Model..............................................................140 Completing Step 1 Â– The Addition of Each Covariate to the Model......143 Step2 Â– Adding Religiosity to the Model...............................................151 Step 3 Â– Testing for Mediational Relationships......................................156 Summary of Findings..........................................................................................161 Chapter Five................................................................................................................... .167 Discussion...........................................................................................................167 Summary of Findings..........................................................................................169 Predicting Depressive Symptomatology in Older Adults.......................169 Predictors of Depressive Symptomatology.............................................177 Predicting Life Satisfaction.....................................................................180 Optimism as a Mediator in the Relati onship Between Religiosity and Life Satisfaction..............................................................................................185 Resilience and Elderly Adults.................................................................187 Limitations of the Study......................................................................................188 Strengths of the Study.........................................................................................191 Implications.........................................................................................................192 Suggestions for Future Research........................................................................194 References..................................................................................................................... ..198 About the Author...................................................................................................End Page
vii List of Tables Table 1. Means, Standard Deviations, Intercor relations, and Coefficient Alpha Reliability.......78 Table 2. Factor Loadings for One-factor Solution of the Religiosity Scale.................................84 Table 3. Correlations Between Individual Items Comprising the Religiosity Index (N = 797)........................................................................................................................... .....85 Table 4. Comparison of the Characteristics of Study Participants with Respondents Who Were Excluded Due to Missing Da ta or Lost to Attrition...........................................109 Table 5. Gender Differences in Descriptive Characteristics of Sample at Baseline (N = 797)........................................................................................................................... ...113 Table 6. ANOVA Results Summarizing Group Differences in Measures Between Included Sample Population and Case s Excluded Due to Missing Data and Between Study Participants and Responde nts Who Were Lost to Attrition................118 Table 7. Comparing the Study Sample with White, Non-Hispanic Adults Aged 65 and Over in Florida.............................................................................................................124 Table 8. Correlations Between Sociodem ographic Variables, Covariates, and the Religiosity at Baseline (N = 797)................................................................................132 Table 9. Correlations Between Control Vari ables, Covariates, Optimism, Time 10 Life Satisfaction and Time 10 Depressi ve Symptomatology (N = 255).............................135
viii Table 10. Summary of ANOVA Results Assess ing the Associations between Categorical Sociodemographic Control Variables, Op timism, Time 10 Life Satisfaction and Time 10 Depressive Symptomatology (N = 255)........................................................137 Table 11. Regression Coefficients Obtained in Multiple Regression Analyses with Control Variables Predicting Time 10 Depressive Symptomatology and Life Satisfaction Â– Model 1 Control Variables (N = 255)...................................................142 Table 12. Building Model 1 Â– Regression Coe fficients and Uniqueness Indices Obtained in Multiple Regression Analyses Pr edicting Depressive Symptomatology Controls + Individual Covariates (N = 255)................................................................144 Table 13. Building Model 1Regression Coeffi cients and Uniqueness Indices Obtained in Multiple Regression Analyses Predicting Life Satisfaction Â– Controls + Individual Covariates...................................................................................................145 Table 14. Model 1 Regression Coefficients Ob tained in Multiple Regression Analyses Predicting Depressive Symptomatology with Controls and Covariates (N = 255).....150 Table 15. Model 2 Predicting Time 10 Depressive Symptomatology and Life Satisfaction Â– Regression Coefficients Obtained in Multiple Regression Analyses with Controls, Covariat es, and Religiosity (N = 255)..................................154 Table 16. Correlational Relationships Am ong Independent, Dependent and Proposed Mediating Variables (N = 255)....................................................................................157 Table 17. Model 3 Testing for Medi ational Relationships (N = 255)......................................160 Table 18. Summary of Regression Analysis fo r Variables Predicting Time 10 Depressive Symptomatology and Time 10 Life Satisfaction (N = 255)........................................163
ix Table 19. Summary Table of Beta Weights of the Most Influential Predictors of Time 10 Life Satisfaction and Time 10 Depr essive Symptomatology (N = 255)......................164
x List of Figures Figure 1. Conceptual Model..................................................................................................... ....13 Figure 2. Satisfaction With Life Scale (SWLS)............................................................................74 Figure 3. Revised CES-D Scale.................................................................................................. ..76 Figure 4. Religiosity Index.................................................................................................... .......82 Figure 5. Life Orientation Test (Optimism)..................................................................................87 Figure 6. PANAS Scale.......................................................................................................... ......93 Figure 7. Sociodemographic Control Vari ables Selected for Study Inclusion.............................95 Figure 8. Variables Selected for Study Inclusion.........................................................................99 Figure 9. Steps in Building the Hierarch ical Regression Model Predicting Time 10 Depressive Symptomatology and Life Satisfaction.....................................................139 Figure 10. Summary of Findings By Study Hypothesis.............................................................165
xi An Investigation of the Relationship Between Religiosity and Subjective Well-being in Older Adults: The Mediating Role of Optimism Teri A. Trede Abstract The proportion of adults aged 65 and older in the United States has been increasing steadily for years. Although most woul d agree that this incr eased longevity is a remarkable achievement, there is growing c oncern regarding the resources that will be necessary to provide care and services fo r this rapidly expanding segment of the population. From a public health perspective, simply increasing longevity is no longer the ultimate outcome sought. A more pertinent goal is increasing the quality of life years, or promoting resilience and successful aging. Stra tegies that assist adults in delaying the onset of disability or reducing the severity of disabling cond itions may serve a vital role in the promotion of resilience and successful aging in older adults. The purpose of this study was to explore the asso ciation between religiosity and subjective well-being, commonly identified as a marker of resilien ce in older adults. In addition, the study sought to determine if that association is mediated by optimism. For this study, subjective well-being was operationalized with measures of depressive symptomatology and life satisfaction. Secondary analyses were conduc ted on longitudinal data collected for the Florida Retirement Study. Re sults indicated that religi osity was not significantly associated with future depressive symptoma tology, but was significantly associated with future life satisfaction. Dispositional optimis m did not mediate the relationships between religiosity and subjective well-being variables.
1 Chapter One Introduction To know how to grow old is the master work of wisdom, and one of the most difficult chapters in the great art of living (Henri Amiel, 1874) Statement of the Problem It is well documented that the United States is undergoing a dramatic demographic shift. The proportion of adults aged 65 and older has been increasing steadily for the past four decades (Coughlin, 1999). A long-term decline in fertil ity rates and enormous improvements in life expectancy have combined to reshape the dem ographic structure of the United States. No longer is our national demographic structure adequately represente d by a pyramid structure, with elderly citizens occupying the smallest, upper segm ents. Rather, the demographic structure now is more adequately represented as a rectangle, with older adult proportions more equal in size to younger segments of the population. This change in demographics clearly reflects the trend of increased longevity that will ultimately redefi ne the demographic map of our country, with elderly citizens emerging as one of the most ra pidly growing segments of our population. With 46% of the population already surviving to at least 80 years of age, and with forecasts predicting a possible rise in life expectancy to 86 in the year 2065 (Lee & Ca rter, 1992), it is important that we focus attention on the health and well-be ing of the increasing elderly population.
2 Conventional definitions of the elderly popul ation generally regard adults between the ages of 65 and 74 as young-old, those 75 to 84 as ol d, and individuals older than 85 as the oldestold. The number of older people in the popul ation is expected to double over the next 20-30 years (Syme, 2003). By 2040, there will be 68 mill ion persons over the age of 65 years of age in the U.S. (Fowler, 1997). The number of persons living to 75 and beyond will grow to 30 million by 2030, and those aged 85 and older will increase to 8 million. In 2015, the Â“Baby Boomers,Â” a cohort of 74 million Americans born between 1946 a nd 1964, will begin to join the ranks of the 65 and over population (Binstock, 1997). It has b een estimated that each day, approximately 6,000 members of that cohort will turn 65, and that by the year 2025, the 65 and older age group will comprise nearly 19% of the total population, a far greater percentage than that of 1900, when those 65 and older accounted for only 4% of the population (Hirsch, 1999; Kerschner & Pegues, 1998). The cohort of adults turning 65 in 2020 is projected to live an average of 19.1 years past their 65th birthday (Lubitz, Beebe, & Baker, 1995). The Graying of America This marked expansion of the 65 and over popul ation presents an even greater imperative when considering that, within that group, those 85 years and older (the ol dest-old), those most likely to need health care supportive services, are the most rapidly growing portion of the population. It has been estimated that this se gment of the population could increase to 8.6 million by the year 2030 and almost double again by the year 2050 (Bauman, Adams, Waldo, 2001; Spencer, 1989; U.S. Census Bureau, 2000). As Siegel (1996) pointed out, this is a projected growth in excess of four hundred percen t, resulting in the oldest old comprising some
3 5% of the total population of the United States in 2050 compared to the 1.4% of the population that they make up today. Most authorities would agree that increased longevity is a remarkable achievement of modern medical science and technology. Billions of dollars have been invested over the past century to improve nutrition, ameliorate public sanitation, advance he alth care and medical technology, prevent epidemics, and reduce diseases and infections. As a result of these efforts, the average lifespan in this country has increas ed from 47 years in 1900, to over 77 years today, and that figure is projected to increase to 83 years by the year 2050 (Rowe & Kahn, 1998). However, though recognized as an extraordinar y achievement, rapid growth of the elderly population also has generated a pprehension regarding the resour ces necessary to provide care and services for these individua ls (Baltes & Carstensen, 1996). Guralnik and colleagues (1996) note that whereas increased longevi ty is significant, it does no t come without consequences. They point to the fact that while many people may escape death due to conditions that often claimed the lives of previous generations, such as infectious diseases, vast numbers of adults may now survive to ages where they suffer from chronic, disabling diseases. They emphasize that as the proportion of adults surviving to very old ages increases, the pu blic health impact and burden of disease and disability, and the ability of older persons to function independently in the community have become critically important is sues. Syme (2003) warns that the system of medical care, already struggling to meet health care demands, will be overwhelmed within the next 40 years as the number of older adults conti nues to increase. Rather than wait for medical care to solve resulting problems after the fact, Syme stresses that we must vigorously explore creative and effective strategies aimed at pr eventing disability a nd promoting health.
4 Other researchers, such as Campion (1998), see the demographic changes as rectangularization of the survival curve, due to the increasing number of deaths occurring after the age of 75. Campion contends that, despite some hesitancy about the quality of life in advanced decades (70s, 80s, and 90s), most indivi duals want to achieve such longevity. Despite medical advancements that make such achievements possible, he cautions that one must avoid casting normal aging as a chronic condition that is managed with expensive medications, endless tests and screenings, and wo rthless interventions. In addition to living longer, larger numbers of persons are retiring at earlier ages. Thus, as life expectancy continues to increase, a grow ing number of individu als enter a period of expanded post-retirement years (Fis her, Mueller, & Cooper, 1991) Â– years when the majority of them will no longer be working for a salary and c ontributing to Social Security. Data from the 2000 Board of Trustees, Federal Hospital Insuran ce Trust Fund illustrates this point: in 1999 there was about one Medicare en rollee for every four workers, with annual expenditures per enrollee running about 18% ($5,400) of the aver age annual wage ($30,200); projections for 2075 show that there will be about one enrollee fo r every two workers, w ith costs per enrollee estimated to be about 24% of th e average annual wage (Miller, 2001). Since life expectancy at age 65 is now 17+ years, in many cases, individual s could be drawing retirement benefits that actually exceed those that they contributed duri ng their employment years. Thus, this rapid growth in the number both of retired persons and of the oldest old has generated much apprehension in regard to the resources and serv ices that these individuals may need as they move through their elder years.
5 Longevity vs. Quality of Life From a public health and aging perspective, simply increasing longevity is no longer the ultimate outcome sought. A more pertinent goal is increasing the quality of life years Â– adding life to years, not just adding years to life, or mo re simply put, promoting successful aging. Fries (1980) suggested that goals rela ted to preventive gerontology should be posed in terms of compression of morbidity compressing the total period of infirmity or disability into a shorter period toward the end of life, rather than me rely expanding the average life span. FriesÂ’ hypothesis challenged the predominant stereotype of old age as a time of decline, helplessness and disability. Instead, he assert ed that healthful living could make it possible for aging adults to survive to an advanced age while maintain ing their vigor and f unctional independence (Campion, 1998; Fries, 1980). Thus, rather th an simply channeling resources into postponing death, a much wiser strategy would be to focus efforts on reducing the period of infirmity before death (Fries, 1980). By concentrating efforts within the compression paradigm the national burden of illness, increasing because of the gr owing number of elderly individuals in the population, may be offset to some extent by a redu ced average illness burden for the individual. This, in turn, could facilitate in creased stability of the health care system (Fries, 2003). Other researchers have not been as optim istic as Fries (1980). The compression of morbidity hypothesis has met with criticism for its navely optimistic ap proach to the impending challenge to the future of our aging populati on. Many demographers and health policy professionals operate from a failure of success paradigm, believing that although medical progress is indeed prolonging life, the resulting months and years co uld be spent in poorer health (Fries, 1980, 2003; Gruenberg, 1977; Rutter & Ru tter, 1993). Similarly, Rutter and Rutter
6 (1993) noted that improvements in overall health are likely to mean that people will live longer, but argued that there does not seem to be much point in prolonging life if it results in decreased independence, declining health, a prolonged vege tative existence and burdensome healthcare costs (p.339). A similarly pessimistic view suggests that medical advances may be more successful in prolonging life than in postponi ng morbidity, resulting not in compression of morbidity, but in expansion of morbidity (Olshansky, Rudberg, Cassel, & Brody, 1991). For example, many individuals in thei r 80s and 90s are subject to se veral life-threatening conditions at once, but delaying the emergence of a single disease for as many as five years in such individuals, disability, suffering and the cost of medical care can be dramatically reduced (Rowe & Kahn, 1998; p.67). Need for the Study Few would argue that we have made enormous strides in lengthening the average lifespan of individuals in our population. However, it could be argued th at we have been less successful when it comes to perfecting strategies to ensure quality of life. As the number of people surviving well beyond the traditi onal retirement age of 65 conti nues to swell, attention will increasingly be turned to concerns about achievi ng increased quality of life and well-being. By examining a range of indicators of subjectiv e well-being and resilience, and elucidating mechanisms that influence such positive outcomes in elders, we can identify factors that have the potential to increase quality of life for older adults. Two increasingly studied determinants of well-being are religious involvement and optimisti c orientation, both of which have been shown to be clearly associated with well-being and de pression. The idea that religion and optimism are linked has long been established in social sciences (Tiger, 199 9). Despite the hypothesized link
7 between them however, few empirical studies have explicitly examined the relationship between them (Mattis, Bontenot, & Hatcher-Kay, 2003). Fewer still have specif ically explored the possibility that optimism may mediate the rela tionship between religiosity and subjective wellbeing indicators (Salsman, Brown, Brechting, & Carlson, 2005). A gr eater understanding of these relationships, and further elucidation of me chanisms of their influence, will likely inspire the development of interventions to promote im proved life satisfaction and quality of life in elderly adults. For example, if it is discove red that religiosity ope rates through optimism to promote positive outcomes and perspectives in older adults, then interventions aimed at instilling a sense of hope and optimism may be beneficial for adults who are less religiously involved or who, due to advancing age and infirmity, find it necessary to disengage from religious activity. In particular, longitudinal resear ch is needed in order to prov ide a clearer understanding of the temporal manner in which these factors operate and relate to one another. Implications for Public Health Both the needs for services and health care expenditures tend to rise with age. However, strategies that assist adults in achieving longevity and delaying the onset of disability or reduce the severity of disabling conditions may increase an individualÂ’s ab ility to remain independent in the community. It seems logical that costs for care would generally be greatest in the last years of life. There are studies, however, that have de monstrated that costs associated with the final year of life actually seem to decline with age. Lubitz, et al., (1995) used Medicare data to determine that in the la st two years of life for persons who died at 101 years of age or older, costs for medical care were 37% ($8,296) of t hose incurred by persons who died at age 70 ($22,590). Miller (2001), also using Medicare data, found that those dying at age 75 incurred
8 costs of $13,500, whereas those dying at age 95 incurred costs of $7,000. Thus, increases in longevity appear to delay the high costs associat ed with the final years of life (Miller, 2001). Aging individuals who are able to remain independent make notably fewer demands upon the health care system (Roos & Haven, 1991) and, in general, demonstrate higher life satisfaction. Additionally, despit e the fact that the elderly acc ount for over one-third of health care spending in the U.S., the average annual Medicare payments are lower for those persons who die at advanced ages (e. g., 80-90 years of age) (Lubitz, et al., 1995). The importance of assisting aging adults in remaini ng independent is clearl y illustrated by the fact that 99% of the population under the age of 75, and 80% of those 85 and older, are residing in the community, not in nursing homes or other long-term care institutions (Hirsch, 1999). Approximately 45 persons per 1000 aged 65 or olde r actually reside in nursing homes (Guttman, 2000). Nursing home costs account for about 20 percent of hea lth care spending for the elderly, and increase with age (Lubitz, et al., 1995). Reducing the need for nursing home utilization is a critical issue, given current projections that Medicare expend itures will outpace revenues prior to the end of the first quarter of the 21st century (Medicare Trust ees, 2003; Roberts, 2003). Thus, as the numbers of aging adults continue s to increase, the elderly represent potential consumers of public health and medical care serv ices in a system already challenged by the increasingly complex needs of a heterogeneous p opulation. Strategies or factors that provide protection against the negative declines and loss es associated with aging, or that promote resiliency and optimal adjustment in the face of the challenges of aging, have the potential to reduce the burden on the public health system and increase the numbers of elders who are able to successfully negotiate the aging process.
9 The State of Gerontological Research on Aging For much of its history, gerontological re search has primarily focused on declines and losses associated with advancing age. Within the last decade several conceptual models have been proposed that have enriched successful aging studies with older adults, an area that was formerly criticized for the lack of theoretical deve lopment. The contribution of such research to our knowledge and understanding of the aging proce ss is immeasurable. However, there remains a lack of consensus among researchers relate d to the maximum potential of the human aging process, the definition of successful aging and appr opriate operationalization of its constructs. It is apparent that old age does indeed encompass losses in physical, cognitive and social domains and, therefore, it is not surp rising that aging is accompanied by anxiety and fear both for individuals (e.g., fear of loss) and societies (e.g., fears of financial burdens) (Baltes & Carstensen, 1996). Such concerns are not withou t merit. However, there is growing evidence that such homogeneous views of aging are unidime nsional and do not take into consideration the possibilities for adaptation, growth, and vitality, and enormous variability in the aging process. Maddox (1987) argues that by focusing on more narrow, theoretically normative outcomes, inadequate attention is paid to the heterogeneity among and within aging adults. Other researchers have been more pointed in their cr iticism, suggesting that aging research needs to include multidimensional and multilevel outcomes th at consider both losses and gains (Baltes & Carstensen, 1996). Inui (2003) contends that su ccessful aging is not adequately understood as mere longevity, and that any attempts to define it must include sufficient well-being in a number of domains (mental, physical, social, spiritual, economic) to sustain th e ability to adapt to changes in oneÂ’s life successfully. Inui stresses the need for a biopsychosocial research approach
10 to the study of successful aging that is driven by transdisciplinary thinking that recognizes the complexity and wide variety of determinants of health and well-bei ng in older adults. Given the growing recognition of the multidim ensional aspects of aging successfully, it seems pertinent to expand the research inquiry beyond asking Â“what is su ccessful aging?Â” to explore the mechanisms through which people co pe with increasing limitations or remain resilient in the face of adversity. Thus, a more germane question may be, Â“ how do people age successfully?Â” For example, more recent gero ntological research suggests that certain individuals demonstrate the ability to maintain or regain normal levels of functioning following adverse life experiences (Staudinger, Marsiske, & Baltes, 1993; Wallace, Bisconti, & Bergeman, 2001). These individuals are though t to possess some combination of internal (e. g., personality and dispositional characteristics, self-esteem, religiosity) and external resources (e. g., social support, social engagement, prosocial behaviors such as voluntary part icipation and health promotion activities) that serve as protective f actors that enhance their resilience and promote more optimal or successful aging. Longitudinal data, in particular, offer the opportunity to identify both characteristics and behavioral patterns of people w ho demonstrate the ability to su ccessfully adapt to, or postpone declines associated with the aging process succes sfully, and to begin to delineate the mechanisms through which they achieve their resilience. Id entifying factors governing resilience in the face of challenges or adversity is an important step in assisting future aging elders in achieving quality of life in their later years. In identi fying resilience factors, it may be particularly advantageous to differentiate between protective f actors that are likely to be innate individual characteristics (such as dispositional characteri stics), and protective mechanisms that may be
11 developed throughout the life span (such as pros ocial behaviors). Such insights will advance knowledge in the field of aging and provide furt her evidence upon which to design interventions aimed at assisting adults in maintaining or deve loping resources to enhance late life resiliency, which can be considered a hallmark of successful aging. Purpose of Study The purpose of this study was twofold. Fi rst, the study was designed to explore the extent to which religious invol vement predicts subjective well -being in elderly adults. The second purpose was to determine the extent to which the association between religiosity and indicators of subjective well-being in elderly adults is mediated by dispositional optimism. There is significant research supporting th e association between sociodemographic, psychological and physical well-being, and dispositio nal traits and successful aging or resilience outcomes. Similar associations have been demo nstrated between prosocial behavior, such as religious involvement, and resilience or successful aging. Determination of causal mechanisms accounting for these relationships has been mo re problematic. There is growing evidence suggesting that there may be more specific mechanisms through which prosocial behaviors actually confer their effects on resiliency and subjectiv e well-being outcomes. Explicating the degree to which these variables contribute to subjective well-being in the elderly may be important for understanding individual differences in more positive outcomes in later life. Conceptual Model The conceptual model for this study (see Figur e 1) was built upon the premise that after controlling for appropriate confounde rs, religiosity would predict greater levels of subjective well-being in elderly adults. It was further pos ited that optimism would mediate the association
12 between religiosity and subjective well-being in elderly adults. In other words, it was hypothesized that, when optimism was added to th e analytic model, not only would a greater proportion of variance be accounted for, but the influence of religiosity on the dependent variables would be reduced, thus providing a more specific explanation of reli giosityÂ’s effects. The study model included as cont rol variables and c ovariates: sociodemographic variables, baseline health and functional ability, and measur es of baseline depressi on, life satisfaction and positive and negative affect. Religiosity was th e independent variable in this study and was measured at Time 1. The dependent variable, subjective well-being, was operationalized as measures of Time 10 life satisfaction and Time 10 depressive symptomatology. Research Questions 1. Is there a positive association between religiosity and subjective well-being in elderly adults? 2. Is the association between re ligiosity and subjective well-bein g in elderly adults mediated by optimism?
13 Time 1 Time10 Conceptual Model Time 5 Sociodemographic Variables Age Gender Marital Status Education Total Assets Living Status Covariates Functional Health Chronic Conditions Health Self-Appraisal Affect Life Satisfaction Depressive Symptomatology Independent Variable Religiosity Mediator Dispositional Optimism Dependent Variables Depressive Symptomatology Life Satisfaction Step 4 Direct Effect Step 7 Step 1 Step 2 Step 3 Step 5 Step 6 Figure 1 Conceptual Model
14 Overview of Study Methods This study was based on secondary analysis of data from an ongoing National Institute on Aging-funded longitudinal study of late-life adaptation to aging conducted by the Elderly Care Research Center of Case Western Reserve University. The study, originally entitled, Buffers of Impairment/Disab ility Cascade Among Old-Old is currently in its 16th year. Initial data collection be gan in 1989 with 1000 adults randomly selected from within a large, age-segreg ated, retirement community in Clearwater, Florida. As a result of eligibility criter ia established for the study, the study sample consists entirely of elderly adults Initial eligibility criteria required that participants be at least 72 years old, reside in Florida at leas t 9 months of the year, and be free of major mental and physical infirmities (Kahana, et al., 2002). The mean age of respondents in the first wave of the study was 79.3 years (Kahana & Kahana, 2002). Respondents are interviewed annually in their homes by a trained interviewer. Each interview utilizes a comprehensive interview instrument that takes between 11/2 and 21/2 hours to complete. The longitudinal nature of this study and the comprehensiveness of the data collection pe rmit the investigation of the effects of prosocial behaviors on resilience outcomes utilizing multidimensional measures. Initial eligibility criteria provided natural controls for a number of factors identified in previous research as possible threats to create spur ious results (e.g., age, functional limitations, lack of residency, income).
15 Research Hypotheses This study tested the following hypotheses: 1. Religiosity is positively associated with s ubjective well-being in elderly adults, after controlling for sociodemographic factors, baseline health and function, and baseline measures of life satisfaction, depressi ve symtomatology and positive and negative affect. 2. The association between reli giosity and subjective well-be ing in elderly adults is mediated by optimism after controlling for so ciodemographic factors, baseline health and function, and baseline measures of life satisfaction and positive and negative affect. Delimitations Delimitations describe the population to which study results may be generalized (Locke, Wyric-Spirduso, & Silverman, 2000). The following delimitations are imposed on this study: 1. Data analysis for this study will be based only on data for residents who reside in one age-segregated Florida retirement community. The vast majority of respondents participating in the waves of data chosen for this study are Caucasian. Although the lack of divers ity in the sample could be seen as reducing generalizability to the population as a whole, the data are generally representative of current resident ial populations of Florida retirement communities.
16 2. Results are only generalizable to elderl y, non-institutionalize d adults and their families who participated in all phases of data collection and had no, or minimal missing data for every variable included in the study. 3. Results are only generalizable to elde rly adults who do not have elevated depressive symptomatology scores prio r to data collection and analysis. 4. Results are generalizable only to White elde rly adults. This is an important point since research has demonstrated that there are numerous differences between White, Black, Hispanic and other minority populations in such areas as health status, functional disabilities, religious involvement, and so cial structure, to name a few. 5. Results are only generalizable to elderl y adults who have minimal functional limitations, including ADL / IADL disabi lities, vision loss or hearing loss, restricting their ability to participate fu lly in activities outside their home. Limitations Limitations refer to limiting conditions or restrictive weaknesses inherent in the study design (Locke et al., 2000). The fo llowing are limitations of this study: 1. Participation in the original study from which the data were drawn was based upon random sampling of a specified comm unity sample. Participation was voluntary, but subject to exclusion crite ria that reduced th e randomness of the final population enrolled in the study. Respondents had to be physically and mentally capable of submitting to a lengthy interview. These requirements may represent threats to the internal validity of the study in that they create selection bias. Thus, those who entered the study were in better health th an those who were
17 excluded, and it may be that more positive outcomes are attributable to the fact that they were in better health to be gin with. The relationships between the independent and dependent variables in the study might be affected by the selectivity of the individuals who are in the sample, or by pre-existing differences between groups in aspects related to the independent variables. 2. Survey results for the original study are based upon self report. Many of the questions contained in the instrument ask for self disclosure of sensitive information. Social desirability bias ha s been found to affect responses on such questionnaires and interviews. Social desirability is the tendency to present oneself in a good light to th e researcher or interviewe r and may provide normative responses or socially acceptable answers to questions rather than truthful answers (Nunnally, 2000). It is also possible that re spondentsÂ’ ratings of measures such as self-appraised health could have been influenced by social comparison bias whereby respondents formulated their re sponses based upon th eir perception of how they compared themselves to their pe ers rather than stri ctly on an objective appraisal of their health or well-being. 3. Participation in the original study may be influenced by the Â“worried wellÂ” phenomenon, whereby healthy, interested persons are more likely to participate in research than persons who may be more si milar to the individuals that are actually targeted by the study. A more heteroge neous sample, one that included subjects who did not represent adults who may alre ady be considered Â“successful agersÂ” would increase generalizability and allo w further extension of the research.
18 4. To increase explanatory power of this st udy, it would have been preferable to include a measure for spirituality. Much current research suggests that spirituality, often included among measures of religiosity, should be considered as a distinctly separate construct that may exert significant influence on the lives of many aging elders Â– not necessarily in c onjunction with religious behaviors or involvement. However, the source data fo r this study did not in clude indicators of this construct in the waves that were analyzed. 5. Although the use of multi-wave panel data for this study allows for the establishment of temporal ordering of se lected indicators and outcomes, one could argue that a complete unders tanding of the true predic tive power of the indicators is impossible without accounting for their e ffects earlier in th e lifespan. Although research demonstrates that dispositiona l optimism and global life satisfaction are relatively stable across the lifespan, it w ould be beneficial to investigate how levels of well-being and life satisfaction earlier in the lifespan influence aging outcomes. For example, it may be that high levels of well-being in early adulthood predict high levels of well-bei ng in old age. Or it may be that wellbeing becomes more significant as one en ters later stages of life when the challenges of aging increase and coping re sources may decline. It may also be that, as elderly adults meet and succe ssfully negotiate the challenges created by the aging process, they e xperience a boost in self-estee m and optimism which, in turn, enhances well-being. Such inves tigation is beyond the scope of this study due to the homogeneous, age restricted nature of the original study sample.
19 Definitions 1. Activities of Daily Living (ADLs) Â– one of the key measures used to determine an individualÂ’s ability to remain indepe ndent. ADLs provide a measure of an individualÂ’s ability to manage their pers onal care and include ba sic functions such as dressing, bathing, toileting, feeding oneself, transferring from bed to chair, and walking. 2. Confounders variables that may exer t influence on either, or both, the independent and dependent variables, include sociodemographic variables, functional health status (ADLs, IADLs), and affect (PA/NA). 3. Depressive Symptomatology symptoms of depression; may or may not be indicative of clinical depression. 4. Functional Limitations Â– impairment in those everyday activities that involve functioning in and out of the home, refe rred to as Activities of Daily Living (ADLs) and Instrumental Activities of Daily Living (IADLs). A person is disabled or dependent when he or sh e cannot perform some of these usual activities without assistance. 5. Instrumental Activities of Daily Living (IADLs) Â– a key measure to assess an individualÂ’s ability to function independe ntly in the community. IADLs assess an individualÂ’s ability to perform non-pers onal care activities such as preparing meals, paying bills, using the telephone, cleaning the house, writing, or reading. 6. Life satisfaction the conscious, cognitively judgmental process individuals use to assess the quality of their lives in accordance with their own unique set of criteria
20 7. Markers of Resilience Â– represented by the global construct of subjective wellbeing operationalized as a respondentÂ’s perception of life satisfaction and depressive symptoms. 8. Optimism refers to an individualÂ’s percep tions of situations and expectations of favorable or positive outcomes over time. 9. Positive and Negative Affectivity (PA/NA) Â– generally considered to represent the emotional components of subjective well-being; two distinguishable dimensions composing mood. Positive affectivity reflects the extent to which individuals generally feel active, alert and enthusiastic, whereas negative affectivity reflects the extent to whic h individuals genera lly feel upset or unpleasantly aroused. 10. Proactive Behaviors Â– behaviors dir ected toward adjustment, coping and adaptation to the challenges older adults f ace as part of the aging process, such as religious involvement, social interactions, etc. 11. Religious Attendance Â– frequency of attend ance of organized religious services. 12. Religiosity Â– a multidimensional construct encompassing religious affiliation, attendance, religious salience, solitary pr ayer, and religious attitudes and beliefs. 13. Religious Salience Â– a measure of privat e religious involvement and beliefs. 14. Resilience Â– describes an older adultÂ’s ability to resist the challenges and forces of decline and deterioration associated with aging to maintain stable health and function; the ability to recover from losse s and illness and regain stability in the face of stresses and challenges.
21 15. Sociodemographic Variables Â– control vari ables including age, gender, marital status, education, total assets. 16. Subjective well-being (SWB) Â– a multidimensional construct that represents an evaluation of how individuals evaluate their lives; includes measures of depressive symptomatology and life satisfaction. 17. Successful Aging Â– not merely the ab sence of disease, but rather, a multidimensional process that reflects optimal adaptation to the challenges, losses and changes presented by the aging proce ss. This optimal adaptation includes the achievement and maintenance of high le vels of subjective well-being, physical and mental health, engagement with lif e, and overall life satisfaction.
22 Chapter Two Background There has been increasing recognition that the proce ss of aging is a complex process involving adaptation to physical, social and psyc hological changes that are intertwined with advancing age. Stev erink and colleagues (1998) suggest that investigations of successful adaptation should incorporate a multi-disciplinary approach that utilizes indicators from multiple life dom ains. Successful aging research has indeed been undertaken in many disciplines and has included indicators from many life domains. However, there remains a lack of consensus about how to define successful aging, and precisely how to operationalize its component s. Thus, comparison across studies has been difficult, if not impossible, and identi fication of mechanisms and processes that contribute to the successful aging of adults has been limited. To develop a multidimensional model that draws upon facets from many fields of study, it is necessary first to explore the c oncepts, constructs and theoretical foundations that contribute to the compone nts of the model. Conceptually, the framework of this study hypothesizes that specific factors, such as religious involvement, predict markers of resilience in older adults. These resilience fact ors are, in turn, believed to be important to the process of successful aging. To provide adequate context fo r the basis of this study, it will be necessary to explore literature relativ e to both successful aging and resilience. To begin, a brief history of theoretical attempts to define and describe both is provided. In
23 addition, literature will be reviewed that de tails the state of curre nt knowledge related to the hypothesized associations among the variable s included in the model for this study. Successful Aging The stereotypical image of older adults has long been a negative one, evoking images of debilitating illness, impaired cogni tive function, forgetfulness, isolation from society and despair. Old age often has been portrayed as a time characterized by multiple losses, deterioration, and lack of purpose and meaning. Fortunately, more positive conceptualizations of aging have become mo re common in the past two to three decades, and much of the fear of old age has begun to be tempered by illustrations of successful negotiation of the aging process. Early Theories Definitions of success vary according to context in which the concept is being utilized. In general usage, success refers to desired achievement, attainment of goals, or as a measure of economic achievement. Some critics contend that successful aging is an oxymoron Â– that aging successfu lly implies not aging at all (Baltes & Carstensen, 1996). Very early theories of successful aging desc ribed highly idealized human states as the ultimate positive outcome of old age. For example, Jung (1931) identified criteria that account for successful aging: expansion be yond gender constraints toward full humanity, life review, psychological interiorization a nd wisdom. Bhler (1933) theorized that successful aging was accomplished with the acceptance of decline as the adaptive task of old age.
24 Later research related to successful ag ing drew direction from Talcott ParsonsÂ’ (1951) Theory of Action, and EriksonÂ’s (1959) life span developmental theory (Ford, Haug, Stange, Gaines, Noelker, & Jones, 2000) Later, using results gathered in the Kansas City Study of Adult Life, Cumming a nd Henry (1961) suggested that, as they reach old age, adults begin to disengage as part of their adjustment to elderly life. They concluded that this withdrawal process was done by mutual agreement between the generations that society, in general, accepte d the notion that older adults should retire and relinquish roles and respons ibilities to make way for younger generations (ie., free up jobs). The theory also assumes that as olde r adults move toward the last years of their life, they no longer desire or requir e engagement with society. The Disengagement Theory was soon cri ticized by other researchers who posited a contrasting view of aging, suggesting that with advancing age, engagement, rather than disengagement, is more closely associated with well-being (Tobin & Neugarten, 1961). Lemon, Bengtson, and Peterson (1972) proposed in the Activity Theory that the key to successful aging is not found in withdrawing from life and shrinking from society, but rather, by maintaining a high level of both physical activity and social integration, findings that were replicated successfully a decade later by Longino and Kart (1982). Knapp (1977) suggested that the theory supports the natural tendency for aging adults to seek interactions with others, particularly in group or community activities. He concluded that older adults who participate in high levels of activity generally express more positive affect, and those who are less engaged expre ss a higher level of negative affect. Thus, older adults who remain active and productive have better outcomes than do those who disengage from society (Gubrium, 1972; Herzog & House, 1991; Lemon, et al., 1972).
25 Activity Theory was used to demonstrate th at activity supplies role support and opportunities for personal devel opment in the elderly (Zimmer, Hickey, & Searle, 1995). For example, evidence has been provided by Coleman and Iso-Ahola (1993) that indicates that socially integrated activities that provide enhanced perceptions of social support and foster companionship act as a buffer against stressful life events. Expansion of Successful Aging Theories Gerontological researcher s have continued to explor e and expand the concept of successful adaptation to old age (Aquino, Russell, Cutrona, & Altmaier, 1996; Bearon, 1996; Havighurst, 1961; Kerchner & Pegues, 1998; Larson, 1978; Lemon, et al., 1972; Palmore, 1979; Roos & Havens, 1981; Tobin & Neugarten, 1961; Wetle, 1997). Although the concept of successful aging was initiated some years ago, the term was popularized in an article published in Science by Rowe and Kahn (1987) (Strawbridge, Wallhagen, & Cohen, 2002) In that article, the researchers advanced the perspective that what many viewed as effects of aging were, in fact, effects of disease. They proposed that individuals aging Â“succe ssfullyÂ” would show little or no physiological age-related decrements, whereas those aging Â“usuallyÂ” would show disease-associated effects of age. This model of successful aging would even tually be broadened as a result of the MacArthur Study of Successful Aging, a 10 year, $10 million study carried out by dozens of scientists focused on developing the con ceptual basis for a new gerontological model encompassing the positive aspects of aging (Rowe & Kahn, 1998). As a result of their work, two additional components were added to the model: maintenance of high physical
26 and cognitive functional capaci ty, and active engagement w ith life (Minkler & Fadem, 2002). In general, much early research describe d successful aging in terms of length of life (Bearon, 1996). Havighurst (1961) defi ned successful aging as Â“adding life to the years,Â” and Â“getting satisfaction from life.Â” As further research evolved in a number of relevant fields, the definition of succ essful aging became more complex and comprehensive. Palmore (1979) explains succ essful aging as a combination of longevity, health, and life satisfaction. Ryff (1982; 1989) proposed a model of successful aging that incorporated developmental, clinical, and mental health pers pectives. She argued that an assessment of success related to aging must consider multiple aspects of life. Ryff identified six dimensions of positive f unctioning in her model: self-acceptance, environmental mastery, positive relations with others, autonomy, personal growth, and purpose in life. Current Views of Successful Aging There also has been increasing consen sus that, although aging is accompanied by a general decline in functional ability, there is a great deal of heterogeneity in this process (Beckett, Brock, Lemke, Mendes, Guralnik et al. 1996; Rowe & Kahn, 1987), as was demonstrated in the Australian Longitudinal Study of Aging (Andrew s, Clark, & Luszcz, 2002). Researchers in this study concluded that not only do people age with differing degrees of success, those aging most successful ly not only live longer, but also report a better quality of life. This variability in th e adaptation to the aging process has fueled the
27 search for factors and characteristics that predict the maintenance of high levels of functioning with age (Michael, et al., 1999). Other attempts to define successful aging have used a variety of adjectives, such as, optimal aging, healthy aging, produc tive aging, hardiness, robust aging, and resilience. In the gerontological literature to date, there is no singl e, universally accepted definition or model of succe ssful aging, or consensus regarding how best to operationalize the concept. Michael and co lleagues (1999) illustra te this lack of consensus with their assertion that most res earch includes one or mo re of four different methods of conceptualizing what they term healthy aging: absence of disease or physiologic decline; high psychological function; high physical function; or some combination of the above. Strawbridge, et al. (2002) take a similar view with their observation that, with few exceptions, most definitions of successf ul aging are still somewhat narrow, placing a strong emphasis on health and physical functioning rather than a more inclusive conceptualization. There is also a lack of consensus as to appropriate approaches to the study of su ccessful aging, what variables should be included in the model, and through which mech anisms is successful aging likely to be achieved. Gains and Losses in Aging As noted above, in recent years, the defi cit approach to aging has been modified and focus has shifted to characterizing the aging process as one of a changing balance between gains and losses, which becomes less positive with advancing age. For example, data from the MacArthur Studies of Successf ul Aging (Seeman, Berkman, Charpentier,
28 Blazer, Albert, & Tinetti, 1995), demonstrate the presence of patterns of both decline and improvement in physical function in old age, suggesting that old age is not inevitably associated with decline. Similarly, research ers have demonstrated that productivity often remains surprisingly high, or even improve s with aging (Glass, Seeman, Herzog, Kahn, & Berkman, 1995). It appears that psychosocial factors play a role in the maintenance of productivity in old age, such as satisfaction with life, mastery, de pression, self efficacy and quality of social support (Larson (1997). More recent models of su ccessful aging have expanded to include an evergrowing array of factors that either contribute to, or serve as barriers to its achievement. Increased attention has been focused on expl oring ways of lengthening the number of years of Â“active life expectancy,Â” and increa sing the quality of life during the later years (Baltes & Lang, 1997; Bearon, 1996). For example, some biomedical researchers have addressed effects of successf ul or optimal aging on morbidity and mortality, and have distinguished Â“successfulÂ” from Â“usualÂ” ag ing (Baltes, & Baltes, 1983; Larson, 1997; Rowe & Kahn, 1987). In this frame of reference, successful agers are those individuals who are able to transcend the normal ag e-related changes and maintain optimal functioning until much later in life. What qualifies as optimal functioning, however, appears to be open to a wide array of in terpretations throughout the literature. Successful Adaptation Aging challenges individuals to act and respond to ever chan ging inner and outer environments. It has been argued that the way that people actively face and adjust to adverse circumstances is what successful aging is all about (Steverink, Lindenberg, &
29 Ormel; 1998). This adaptivity, or behavioral plasticity, is seen as an important criterion for achieving successful aging (Baltes & Baltes, 1990). In this conceptual framework, successful aging resides in an individualÂ’s ability to use selective optimization with compensation (SOC) successfully to adapt to the biological and psychological changes that occur with the aging process. The framework is useful in the understanding of resilience across the lif e span (Baltes & Carstensen, 1996; Baltes, 1997; Freund & Baltes, 1998). The process of selective optimization with compensation can be seen as a lifelong process of adaptation that involves the dynami c interplay between gain s and losses, and is based upon the assumption that individuals orch estrate adaptive res ources reflective of the model components (selection, optimizati on, and compensation) as they respond to both developmental opportunities, such as educ ation, or limitations such as illness. As the ratio of gains and losses becomes increa singly negative a greater demand for adaptive responses to the more dele terious changes is created (Carstensen & Freund, 1994; Heckhausen, Dixon, & Baltes, 1989). As older adults encounter losses, such as those in biological, mental or social reserves, they re spond by activating one or more of the model component processes Â– selecti on, optimization, or compensa tion (Baltes & Baltes, 1990; Carstensen & Freund, 1994; Freund & Baltes, 1998). When implemented together, use of the processes enables people to achieve their goals despite, or even because of, losses and increasing vulnerabilities associated w ith aging (Baltes & Carstensen, 1996). Baltes and Carstensen (1996) contend that focusing on a reduction in reserves and an increasing number of specific losses a nd challenges in the biological, social, and psychological spheres on one side, and potential for growth and plasticity in old age on the other side, successful aging can best be understood. Such a model, they argue, can
30 accommodate a great diversity of outcomes, allow for different success criteria, and explain how people achieve successful agi ng in the face of simultaneous losses in multiple life domains. Thus, rather than de ny the inevitable losses that face all elderly individuals in advanced age, the SOC model proposes that old age holds the potential to be a time when the accumulated knowledge and experience of a life-time can be applied to the realization of new, mean ingful goals and pursuits. Alternative Models An alternative model of successful adap tation has been suggested by Heckhausen and Schulz (1993), who extended and elaborat ed the SOC model by integrating it with their life-span theory of control (Heckhausen & Schulz, 1995; Schulz & Heckhausen, 1996). Their model of developmental regula tion across the life course assumes that individuals are confronted w ith varying challenges to th e maintenance of primary and secondary control throughout the life span. Prim ary control refers to attempts to change the world so that it meets th e needs and desires of the in dividual. Secondary control includes attempts to bring oneself in line wi th the environment. Changes in biological and societal potentials and cons traints over the life course alte r an individualÂ’s potential for control, and successful adaptation, or ba lance between primary and secondary control, is dependent upon the developmental potential of the individual in th e different life stages (Steverink, et al., 1998) as well as effectiv e use of selection and compensation. The effective use of control strategies is also regulated by optimization.
31 Yet another model proposes th at older adults experience considerable strain as they go through the aging proce ss, and that successful outc omes are dependent upon the use of three coping strategies necessary to adapt to the changes and discontinuities of aging and maintain a positive sense of self in old age Â– assimilation, accommodation and immunization (Brandstdter & Greve, 1994; Brandstdter & Renner, 1990; Brandstdter, Wentura, & Greve, 1993). Assimilation and accommodation processes are used to help maintain a sense of control and positive vi ew of self and pers onal development. Immunizing mechanisms influence the processing of self-relevant information so that an individualÂ’s self-descriptions are immunized against discre pant evidence (Brandstdter & Greve, (1994). Providing empirical evidence to illustrate their assumptions, Brandstdter and Greve (1994) contend that successful agi ng is reflected by an age-related shift from assimilative to accommodative modes of coping over the life course. Research Gaps It has been argued that studies that have illustrated that the aging process involves losses and gains, and limits as well as potentia ls (Baltes & Carstensen 1996; Scheidt, et al., 1999), help to generate questions about the meaning and signifi cance of aging and old age. But further research is needed to de velop theoretical conceptions that move beyond generalized dichotomous models of successfu l/unsuccessful or heal thy/unhealthy aging (Minkler & Fadem, 2002). Despite consider able growth in knowledge and understanding of the aging process, there re main many questions related to successful aging. As noted previously, there is still a lack of consensus regarding the appropriate definition of successful aging and how best to operationali ze the concept. Until these definitional and
32 operationalizing issues are resolved, empirical comparisons across studies to definitively construct appropriate explanatory models of successful aging will be hampered. Inui (2003) has argued that any attempts to desc ribe, understand or promote successful aging must include transdisciplinary thinking that recognizes the complexity and multidimensionality of determinants of h ealth in aging adults. In line with a transdisciplinary research approach, Inui s uggests that the foundation for successful aging can be found among the various states of wellbeing, including biological, psychological, social, spiritual, economic, and medical. In his view, a rich template for conceptualization and measurement is necessa ry to identify some of what he terms Â“higher-orderÂ” patterns th at distinguish successful ag ing and resilience. Resilience Defining Resilience Another body of research that has releva nce to the study of successful aging is work exploring resilience in individuals. Th e term resilience has been used in a variety of perspectives and given a multitude of definitions. For example, resilience has been used to describe an older individualÂ’s abil ity to cope successfully with the inevitable challenges and changes of the aging pro cess (e.g., Brandstdter & Greve, 1994; Carstensen & Freund, 1994; Felten, 2000; J acelon, 1997; Moen & Erickson, 1995), or the ability of some people to resist or reject the potential negative effects caused by adversity or risk factors that make people vulnerabl e or undermine their well-being (Bachay & Cingel, 1999). The term resilience has al so has been used interchangeably with hardiness, a practice that Felton (2000) cont ends has obscured the conceptual meaning of
33 resilience. Although the terms seem somewhat similar in meaning, they convey distinctly different responses, particularly in an elde rly population (Felton, 2000). Conceptually, the term hardiness is used appropriately to indicate resistance to illness or adaptation to stress (Kobasa, 1979; Selye, 1956), whereas re silience is a respons e demonstrating not just rebound, but also includes improveme nt of function following a negative circumstance (Felton, 2000). Initial Resilience Research The initial ground work of resilience re search dates back to the 1950s and 1960s, in work by Emmy Werner, Michael Rutter, and Norman Garmezy (Werner & Smith, 1977). Most of the early work related to resilience focuses on children and adolescents (e.g., Beardslee & Podorefsky, 1988; Cowen & Work, 1988; Garmezy, 1985; 1991; 1993); Masten & OÂ’Connor, 1989; Mulholland, Watt, Philpott, & Sarlin, 1991). Researchers in childhood and adolescent deve lopment identified resilience as a factor enabling children to develop successfully in to well-adjusted adults despite exposure to adverse circumstances early in life (Hardy, Concato, & Gill, 2004). Rew and colleagues (2001) concluded that homeless youth who perceived themselves as resilient, even though disconnected from other people, were less lonely, less hopeless, and engaged in fewer life-threatening behaviors. Resilience and Adults Resilience research with adults was or iginally focused on outcomes, such as psychosocial function following the death of a spouse (McCr ae & Costa, 1988), or on the basis of personality traits that moderate a stress response (Hardy, Concato, & Gill, 2004).
34 More recently, the term has been used to de scribe a process that incorporates certain internal traits and external factors to surviv e adverse circumstances or cope with stress (Richardson, 2002; Hardy, et al., 2004). Resilience research with adults has been pe rformed in a variety of disciplines, and has included psychological, sociological, biol ogical, and multidimensional constructs. Studies have focused on resilience in the face of devastating illnesses, such as cancer or AIDS (Antoni & Goodkin, 1988; Rabkin, Re mien, Katoff, & Williams, 1993), or have used resilience to explain individuals who a ppear more capable of overcoming stressful situations, negative life events, negative environments, or adjusting to traumatic losses (Aldwin, Levenson, & Spiro, III, 1994; Ru tter, 1985; Solomon & Prager, 1992). Still others have looked at resilience as a protec tive aspect in the study of successful aging (Rutter, 1987; Wagnild & Young, 1990). Bauman et al. (2002) reviewed several studies that focused on a wide array resilience perspec tives. They called attention to the fact that the research illustrates a continued lack of consistency in how the term resilience is interpreted. For example, se veral researchers have described resilience as a return to functional levels present prior to adve rse circumstances (Rowe & Kahn 1997; Rutter 1987; Staudinger & Fleeson, 1996). Others ha ve argued that resilience is demonstrated by improvement or growth beyond previous levels following adversity (Felton, 2000; Foster, 1997; LeFerriere & Hamel-Bissell, 1994; Polk, 1997; Wagnild & Young, 1990). More recently, resilience research has been extended to include other age groups in the life span, particularly older adults (Bauman, Adams & Waldo, 2000; Brandtstdter & Greve, 1994; Felton, 2000; Hardy, et al ., 2004; Jacelon, 1997; Liem, James, OÂ’Toole, & Boudewyn, 1997; Luthar & Cicchetti, 2000; Moen & Erickson, 1995; Rabkin, Remien,
35 Katoff, & Williams, 1993; Wagnild & Y oung, 1990; 1993). Current gerontological research suggests that resiliency is a dynamic process that involves protective factors that encompass the physical, psychological, social, an d spiritual domains of older adultsÂ’ lives (Nakashima & Canda, 2005; Staudinger, Mars iske, & Baltes, 1995). Some researchers have argued that there is a need to move beyond focusing on decline and deterioration in the elderly and recognize potential for st ability and even improvement in old age (Beckett, Brock, Lemke, Mendes, Guralnik et al., 1996; Crimmins, Saito, & Reynolds, 1997; Strawbridge, Kaplan, Camacho, & Cohen, 1992). Responses to the challenges of the old age (illness, social isolation, physical disability, functional decline) influence the trajectory of the aging process. Those who are more vulnerable to these challenges, who decline or die, are conceptually different from those who are able to resist the challenges and maintain stable health and function, or even succumb temporarily but then recover (Walter-Ginzburg, Shmotkin, Blumstein, & Shor ek, 2005). This resistance to the forces of decline and deterioration, or stability in the face of stresses and challenges can offset losses. Taken together, stability and impr ovement define resili ence (Brandstdter & Greve, 1994; Gelten, 2000; Walter-Ginzburg, et al., 2005). Regardless of which definition is refe renced, it appears th at many of them encompass common themes: Resilience surfaces in the face of hardsh ip and involves the way in which individuals respond to challenges; individuals exhibiti ng resilience are able to bounce back or rebound from adversity, re aching or surpassing a pre-crisis level of functioning; and resilience is often described in terms of wellness or strengths, rather than pathology or deficits (Hawley, 2000; Hawley & DeHann, 1996).
36 Protective Factors Related to Resilience Resilience is often discussed in terms of protective factors that may serve as resources to buffer the effects of challenges or adversity (Hawley, 2000). Resilience is more likely to be present in situations wh ere factors are minimized by the presence of protective factors (Hawley, 2000). Protec tive factors might include dispositional characteristics or temperament (Pitkala, Laakkonen, Strandberg, & Tilvis, 2004; Siegler & Brummett, 2000), social resources (Holahan & Holahan, 1987), or re ligiosity (Martin, 2002). Some researchers have integrated re silience as a construct in their conceptual framework, and argue that it represents: (a ) stress and risk mechanisms, (b) coping and protective mechanisms, or (c) buffers or moderating mechanisms (Egeland, Carlson, & Sroufe, 1993; Rew, et al., 2001; Rutter, 1987). Rutter (1987) contends that people are able to adapt and cope with adversity or ne gative events as long as they are able to balance risks, stressful life events, and pr otective factors adequately. Considering the heterogeneity of the aging population, it is likely that protectiv e factors vary among individuals. Werner (1990) suggests that the concepts of resilience and protective factors are the obverse of vulnerability a nd risk factors. Other resear chers have asserted that two sets of protective factors promote resilience when confronted with challenges, social resources and personal resources (Moen & Erickson, 1995; Rutter, 1987; Schaefer & Moos, 1992). The identification of specific factors that offer a protective influence against risk factors associated with aging coul d assist in the development of interventions aimed at increasing resilience in the elderly. As with other areas of resilience research, there is a common focus on the multidimensionality of the conceptual framework. Sperry (1992) defines resilience as a multidimen sional construct encompassing emotional,
37 spiritual, social, cognitive, and physical domains. Bauman and colleagues (2001) concur with SperryÂ’s view that these domains are interdependent, and thus suggest that they should not be considered in isolation. Rath er, resilience should be explored as a holistic construct. Resilient Adults Resilient individuals are more likely to possess some combination of internal (e.g., personality or dispositional characteris tics) and external (e .g., social support) resources which to draw when faced with adversity (Wallace, Bisconti, & Bergeman, 2001). Resilient elders tend to maintain a positive sense of well-being in the face of loss of spouse and friends, loss of physical functi on, and decline of soci al status that are commonly a part of the aging process (Bau man, et al., 2001; Carstensen & Freund, 1994), and are better able to recover from th e impact of difficult circumstances, such as difficult transitions or crises (Bauman, et al 2001). Resilient individuals have also been described as exhibiting social and psychological competence characterized by five major themes: (equanimity, or having a balanced pe rspective of oneÂ’s life experience; (2) perseverance, or being able to carry on in sp ite of adversity, and a willingness to continue the struggle; (3) self-reliance, or a be lief in oneself and oneÂ’s abilities; (4) meaningfulness, or the realization that life has a purpose, and that one can contribute meaningfully; (5) existential aloneness, or the realization that the path that each individual takes is unique (Wagnild & Young, 1993). In the oldest old, resilience has been meas ured as a return to previous levels of functioning post adversity, and some researcher s have suggested that such a process may
38 actually lead to growth and improved functioning (Bauman et. al., 2002; Dyer & McGuinness, 1996; LaFerriere & Hamel-Bissell, 1994). This distinction is important in conjunction with older adults because their ability not only to rebound but improve in function after a devastating illness, is a critical ch aracteristic if they are to continue to live independently (Felton, 2000). How individuals are able to cope with st ressful experiences is often determined by the availability of individual and social res ources (Krause, 1997; Ma rtin, 2002). A stable repertoire of resources can help individuals to compensate for the losses and adjustments that accompany the aging process (Marti n, 2002; Steverink, Westerhof, Bode, & Dittmann-Kohli, 2001). For example, disposi tional factors that underlie resiliency in aging adults may enhance their ability to cope with adversity. Identifying these factors may assist service providers in predicting whic h adults are at greatest risk of succumbing to negative circumstances and which adults have the potential to resist prolonged negative consequences of adversity successfully (Jacelon, 1997). This understanding will enable professionals to identify which adults could benefit from encouragement to use their skills to persevere, and which adults should be assisted in the development of skills to help them negotiate the effects of variations in health (Jacelon, 1997). Resilience assessment also will improve cliniciansÂ’ ability to better predict which older adults will be able to return to previous functional levels successfully following acute illness or adverse situation and which individuals may benefit from more customized intervention (Hardy, Concato, & Gill, 2004).
39 Reserve Resources and Resilience Outcomes It has been suggested that the most optimal outcomes are achieved through a combination of protective factors working together (Wallace, et. al., 2001). For example, individual resources work in tandem with various family/community support factors, such as social support, to provide protective effects or resilience out comes in older adults (Bergeman & Wallace, 1999; Garmezy, 1985; Rutter, 1987; Wallace, et. al., 2001). Resilience mechanisms appear to serve as im portant potential reserv e resources that can be drawn upon to compensate for increased losses and change associated with aging (Wallace, et al. 2001). Wallace and colleague s have argued that this relationship is consistent with life-span developmental perspe ctive. They contend that the possibility that more optimal outcomes result from the interplay between resilience resources, is in line with the conceptual notion of intra-indivi dual plasticity, or within-person variability, in development. The researchers note that whereas these resources do not protect an individual from loss or decline, they do enha nce ability to maintain, regain, or advance beyond the level of functioning that precedes ad verse experiences. Thus, these resilience resources could be described as Â“latent rese rvesÂ” on which an indivi dual may draw when faced with adversity. Purpose of Study Building upon previous research that has suggested an association between religious involvement and a number of resilien ce factors associated with successful aging in elderly adults, this study explored the extent to which religiosity predicts subjective well-being in a population of elderly adults residing in an age-segregated retirement
40 community in Florida. Additionally, the st udy investigated the possibility that the relationship between religious involvement and subjective well-being in elderly adults is mediated by optimism, a factor that has been linked similarly to more positive outcomes in elderly adults. Dependent Variables Â– Subjective Well-being Identifying and Predicting Subjective Well-being Conceptually, the dependent variables in this study are constructs that have been used to represent markers of resilience and suc cessful aging as suggest ed in the literature, and are operationalized as measures of s ubjective well-being. The variables, life satisfaction and depressive symptomatology, were chosen to represent a bidimensional measurement of subjective well-being. Th e independent variab le, religiosity, was measured with an index that included severa l aspects of religious involvement suggested by other researchers as important elements in the association between religion and positive outcomes in individuals. Optimism, the proposed mediating variable, was chosen based on accumulating evidence that suggests that positive life orientation, or optimism, predicts favorable outcomes afte r controlling for possible confounders (Chang & Sanna, 2001; Pitkala, Laakonen, Stra ndberg, & Tilvis, 2004) and its proposed relationship to activity and social in teractions (Pitkala et al., 2004). Subjective Well-being Subjective well-being (SWB) is generally used to refer to how people evaluate their lives, and may include such variables as life satisfaction, lack of depression and anxiety and positive moods and emotions (Diener & Emmons, 1984; Pavot & Diener,
41 1993). For example, a person might be said to have high (SWB) if he or she experiences life satisfaction, frequently experiences posit ive emotions and infrequently experiences debilitating sadness or depression. Because subjective well-being is one of the major outcome variables in the field of aging, ther e are a number of measures available (e.g., Bradburn, 1969; Lawton, 1975; Neugarten, Ha vighurst, & Tobin, 1961; Watson, Clark, & Tellegen, 1988). Subjective well-being also has been found to be correlated with optimism (Scheier & Carver, 1985), self-e steem (Wilson, 1967), and control beliefs (Lachman & Weaver, 1998). Individuals with more positive self-i llusions, including unrealistically positive self-perceptions or ove rly optimistic views of the future, not only have higher SWB but other qualities such as ca ring for others and the ability to engage in productive work (Taylor & Brown, 1988). Th ese positive illusions and higher SWB, in turn, are correlated with successful adjust ment to stressful circumstances, including extreme adversity (Taylor & Armor, 1996). High SWB has been used to imply a happy life with many pleasant, and few unpleasan t, experiences and high life satisfaction (Schimmack, Oishi, Furr, & Funder, 2004). For this study, subjective well-being was operationalized as a bidimensional construct and measured with scales representing life satisfaction and depressive symptomatology. Life Satisfaction Moods and emotions, labeled affect, repres ent peopleÂ’s evaluations of the events that occur in their lives (Die ner, et al., 1999). Life satisfaction forms a separate factor from the two major types of affect (PA/ NA) (Andrews & Withey, 1976; Lucas, Diener, & Suh, 1996). Life satisfaction refers to the consci ous, cognitively judgmental process
42 individuals use to assess the quality of thei r lives in accordance with their own unique set of criteria (Shin & Johnson, 1978). Diener and colleagues (1985) argue that, because individuals tend to assign di fferent standards of success to different domains of their lives, it is necessary to assess an individualÂ’s global judgment of his or her life, rather than focusing solely on satisfaction with specific domains. Depressive Symptomatology Depression is a common and often serious clinical condition that is especially prevalent in older individuals with acute a nd chronic medical condi tions (Feinson, 1985; Lewinsohn, Seeley, Roberts, & Allen, 1997; Waxman, Carner, & Klein, 1984). Community-based epidemiologic studies have shown an association between depressive symptoms and mortality (Zheng, Macera, Crof t, Giles, Davis, & Scott, 1997), cognitive deficits such as disturbances in thi nking, problem solving and self-efficacy, and diminished adaptive skills in the face of illness (Covinsky, Kahana, Chin, Palmer, Fortinsky, & Landefeld, 1999). Depression also has been associated with feelings of hopelessness, low self-esteem and sadness, d ecreased activity that may contribute to functional loss, and decreased ability to marshal social s upport and assistance (Covinsky, et al., 1999). These processes, in turn, may accelerate functional decline and prevent older adults from successfully negotiating ch allenges associated with chronic illness (Bruce, Seeman, Merrill, & Blazer, 1994) or dec lines associated with the aging process. Depressive symptomatology might also indi cate a loss of sense of meaning in life (Rothermund & Brandstadter, 2003). From a resilience perspective, however, older adults may be able to activate other resour ces to maximize the quality and quantity of
43 oneÂ’s life (M. M. Baltes & Lang, 1997), t hus reducing the effects of depression. Resources such as religion (Ellison, 1994) and life orientatio ns (Brandtstadter & Rothermund, 2002) may help aging individuals to derive meaning from other domains of life. Independent Variables Resilience and Proactive Behaviors To understand the aging process more comp letely, it is critical to recognize that aging takes place in a social context (Albert, Im, & Raveis, 2002). Findings from recent gerontological research illustrate that so cial and psychological environments play important roles in disability in old age. For example, older people who experience physical and cognitive impairments often alter their environments in ways that enable them to prolong their independence (Albert, et al., 2002). Thus, when studying older adults, it is important to identify psychologica l, social, and environmental characteristics or behaviors that promote resiliency or enhance adjustment to facilitate survival into later ages. Sociological views of successful aging emphasize proactive skills such as coping with stress and loss, the ability to reframe ne gative life experiences, and social interaction as means of achieving successful adaptation to the aging process (Ford, et al., 2000). Kahana and colleagues (1996; 2002; 2003) propo se a model that features prosocial activities as mediators or moderators to st ressful situations to demonstrate successful adaptation in the elderly. Their multi-fa ctorial model, titled Proactive Options for Successful Aging (Kahana, et al., 2002), pl aces an emphasis on proactive/preventive
44 behaviors (e. g., health promotion, planning, helping others) and corrective adaptations (e. g., marshalling support, ro le substitution, environmenta l modification) as means to cope with challenges and losses associated with aging, as well as focuses on social interactions as factors which he lp older adults achieve successful aging. In this context, proactive behaviors act to forestall or dela y the emergence of stressors common to late life, such as chronic illnesses, and successful aging is characterized as positive affective states, a sense of meaning in lif e, and maintenance of valued activities and relationships. In the current study, prosocial behavior was op erationalized as religi osity. To reduce the possibility of spurious associations creat ed by the influence of certain dispositional characteristics on religiosity, positive and nega tive affect, each likely to be correlated with the independent variable, were include d in the model as covariates. Proactive behaviors, incl uding leisure, religious a nd productive activities often take place within the context of social relationships. Numerous studies have demonstrated associations between interper sonal relationships and lower mortality and improved subjective health and well-being (Berkman & Syme, 1979; Blazer, 1982; House, Robbins, & Metzner, 1982; Orth-Gomr, Rosengren, & Wilhelmsen, 1993; Seeman, Kaplan, Knudsen, Cohen, & Guraln ik, 1987; Vogt, Mullooly, Ernst, Pope, & Hollis, 1992). Religious activities are likely to produce interpersonal relationships and increase the range of informal social res ources available to older adults (Midlarsky & Kahana, 1994; Pearlin, 2001; Rosenberg & Mc Cullough, 1981). Research has suggested that older adults who have a greater degree of social involvement generally tend to have lower mortality rates than those who are less involved (Seeman, et al., 1987). However, social interactions often take place in the c ontext of other types of activities, such as
45 religious participation. As a result, it has fre quently been difficult for researchers to sort out the causal path of variable s, that is, to distinguish wh ich benefits are conveyed by the social aspect, and which derive from the ac tual content of the activities (Lennartsson & Silverstein, 2001). It is also possible th at the selection hypothesis might provide explanations for a number of associations between activities and such positive outcomes as reduced mortality and enhanced health. In other words, it may be the case that those adults who have better health and are more active to begin with are the ones that are experiencing reduced mortality and enhanced health, regardless of their participation in these activities. Religiosity Prevalence of Religious Behavior and Affiliation Religiosity and spirituality are distinctly different constructs that constitute integral forces in the lives of many individuals. It is estimated that as many as 90% of American adults report some manner of formal religious affiliation (Kosmin & Lachman, 1993). Gallup (1995) polls indicate that near ly 96% of Americans believe in God or a universal spirit, 42% report regular worship service attendance, 60% indicate that religion is Â“very importantÂ” to them (McCullough, Larson, Hoyt, Koenig, & Thoresen, 2000), and approximately 88% report that they pray to God (Hoge, 1996). There is a substantial literature that connects religion and spiritua lity to physical hea lth (Hill & Pargament, 2003; Powell, Shahabi, & Thoresen, 2003; Seeman, Dubin, & Seeman, 2003; Thoresen, 1999) and mental health and well-being (e.g Donahue, 1985; Krause, 1992; Idler, 1987). Studies from a number of disciplines ge nerally have found a salutary relationship
46 between religious involvemen t and a broad range of hea lth outcomes (Chatters, 2000; Ellison, 1991; Idler, 1987; Levin & Va nderpool, 1989; Musick, 1996; Oman & Reed, 1998), as well as lower mortality (Hummer, Rogers, Nam, & Ellison, 1999; Kark, Shemi, Friedlander, Martin, Manor, & Blondheim, 1996; McCullough, et al., 2000; Oxman, Freeman, & Manheimer, 1995; Strawbridge, Cohen, Shema, & Kaplan, 1997). Moreover, many Americans report believing th at there is a connect ion between religion and health, with 79% reporting their belief th at God answers prayers for healing from incurable diseases (Woodward, 1997), and 25% reporting that they pray regularly to be healed from illness ( Eisenberg, Kessler, Foster, Norlock, Calkins, & Delbanco, 1993). Levin (1994) reviewed over 200 published empirical reports from medicine and epidemiology, and reported the existence of statistically signifi cant linkages between religion and such health conditions as, cardi ovascular disease, stroke, hypertension, cancer, as well as self-rated health. Those w ho attend religious services more often have been shown to be less likely to be admitted to the hospital, have fewer admissions overall, and spend fewer days there than those with less religious affiliation (Koenig & Larson, 1998). However, there also have been contradictory results produced by researchers who found no association between religious coping and hos pital use (Koenig, 1995; Levin & Markides, 1985). Religion and Well-being Religiosity, as an individual resource, can be defined as personal beliefs and behaviors that express a se nse of relatedness to something greater than the self (Wilkinson & Lynn, 2001). Religious involveme nt has been related to better physical
47 functioning, subjective appraisa l of health, life satisfacti on, well-being, and personal adjustment (Neill & Kahn, 1999). Studies demonstrating the positive relationship between religiosity and an arra y of indicators of well-being in older adults date back to the 1940Â’s (Campbell, 1975; Hunsberger, 1985; Id ler, Musick, Ellison, George, Krause, et al., 2003; Koenig, Kvale, & Ferrel, 1988; Krause, 1998; Larson, 1978; McFadden, 1995; Moberg, 1990; Okun & Stock, 1987). More religious individuals may be able to draw on beliefs and traditions of their faith to allay fear, gain hope or put events into perspective (Hadaway & Roof, 1978; Idler, et al., 2003; Reker, Peacock, & Wong, 1987). When faced with physical declines associated with aging, religious beliefs may be an important source of hope and comfort, and may help aging adults to cope with stress, disability and the loss es of loved ones (Koenig, George & Siegler, 1988). Religiously oriented individuals may find more meaning or purpose in life (Musick, Traphagan, Koenig, and Larson, 2000), which in turn im pacts their attitude, act ivities and well-being (McFadden, 1995). Elder adults may experien ce numerous difficulties such as ill-health, reduction of income, and a shrinking social ne twork but still retain a sense of well-being and the conviction that life has meaning. In some elder adults, this demonstration of resilience may be linked with religious commitments that not only affect subjective assessments of well-being, but may also infl uence objective indicators such as health and social support (Dull & Skokan, 1995; Ge orge & Clipp, 1991; McFadden, 1995). Research has demonstrated that amo ng older adults, religious beliefs and practices are associated with higher self-esteem and feelings of self worth (Krause, 1995), and that those who reporting a strong religious faith describe themselves as happier and more satisfied with their lives (Ellison, 1991). Religious coping strategies appear to offer
48 particular protective effects, especially in s ituations of bereavemen t or serious illness, over which an individual may have little direct control (Mattlin, Wetherington, & Kessler, 1990). Such situations are comm on challenges faced by elderly individuals. Religious belief systems offer resources for unde rstanding tragic or stressful events and often can enable individuals to come to te rms with such situations and enhance their ability to find acceptance and peace (Ell ison, 1991; Idler, et al., 2003). There appears to be little consensus in the research as to which specific dimensions of religiosity account for the asso ciations with well-bei ng. The literature indicates that religion may enhance various di mensions of well-being in at least four ways: 1) through social support and integrati on; 2) through the establishment of personal relationships with a divine other; 3) through the provision of systems of meaning and existential coherence; and 4) through the prom otion of more specific patterns of religious organization and personal lifestyle (Ellison, 1991). Age Effects on Religious Involvement Although levels of religious attendance ma y decrease, levels of religiosity among individuals tend to increase as they age (Blazer & Palmore, 1976). Empirical data indicate that religion remains an important factor in the lives of elderly adults (Commerford & Reznikoff, 1996). Older people often turn to reli gion as a means of coping with stressful events or crises (Koe nig, George, & Siegler, 1988; McCrae, 1984). Coping strategies might include praying, s eeking spiritual suppor t and guidance, or seeking social support from the religious co mmunity (Pargament, Ensing, Falgout, Olsen, Reilly, et al., 1990).
49 Not all studies, however, have demonstrated conclusive evidence for the salutary relationship between higher levels of religious involvement and positive outcomes (Benjamins, 2004; Markides, Levin, & Ray, 1987) For example, Markides, et al. (1987) found that when dropouts from successive wave s of an 8-year study were excluded from analyses, the relationship between religious attendance and life satisfaction disappeared. Additionally, Benjamins (2004) ca lls attention to gaps in the research. She notes that only a small number of studies have concentrated on functional limitations. A large proportion of elderly adults are challenged by functional limitations, and these limitations often prevent the performance of everyday ac tivities, reduce mobility, and can lead to a loss of independence. The studies that have addressed the influence of religious factors on functional limitations have produced conflicti ng results as to the association between the two variables (Idler & Kasl, 1992; 1997; Krause, 1998), with one set of studies finding that religious attendan ce predicts higher levels of physical functioning, and the other finding no association between religious variables and functional disability. Possible Mechanisms of ReligionÂ’s Influence Religion serves as both a social instituti on and a source of exis tential meaning to individuals, and as such, is an important person al resource for older ad ults in terms of the provision of services and fellowship and a ssistance with coping and adaptation with issues of daily life, change, loss and deat h (Levin, 1994). Epidemiological findings have demonstrated that as adults age religious involvement becomes more strongly predictive of better health, happiness and life satisfaction. Levin ( 1994) notes that more private expressions of religiosity and positive subjec tive religious attitudes also tend to promote
50 health, especially when factor s such as declining health, re strict more formal and public participation. McFadden (1995) similarly sugge sts that more solitary religious activities, such as prayer and meditation may produce positive effects, such as emotional calm. Hypotheses proposed to explain the mechanisms through which religion influences health include, the health prom oting behaviors advocated by some religions (Cochran, Beeghley, & Bock, 1994; Ellison, 1994) the social support available in religious communities (Ellison, 1991), the subjective perceptions of well-being effectuated by religious beliefs (Levin 1994a 1994b), and the emotional comfort created by religious ritual (Levin, 1994a, 1994b). Th ese hypotheses have been applied to studies of both physical and mental well-being (McFa dden, 1995). Some research has indicated that religion is the coping resource most frequently c ited by older adults (Koenig, George, & Siegler, 1988) and suggests that re ligionÂ’s stress buffering properties affect how older adults experience both physical illness and mental distress, perhaps by compensating for the erosion of feelings of mastery and self-esteem produced by stress (Krause & Tran, 1989). Ellison (1994) also sugg ests that religion plays a significant role in the stress paradigm, serving to impede de pression by reducing the risk of certain major stressors, generating relatively high levels of objective and subjective social resources, and enhancing positive self-perceptions and other psychological resources. Thus, since older age may bring about more challenges and shrinking of resources, the resulting higher levels of stress may increase the infl uence that religion may have on health (Krause & Tran, 1989) Many researchers have suggested that positive aspects of religion may be experienced as a result of religious philo sophy or teachings in that many health
51 promoting or health threatening behaviors may be prescribed or proscribed by particular religions or religious denominations (Ellis on, 1991; Gardner & Lyon, 1982; Idler, 1987; Levin, 1994a, 1994b; Levin & Vanderpool, 1989). Additionally, religious individuals may have religioethical norms that guide th eir behaviors (Ellison & Levin, 1998). These governances may lead individuals to abstai n from unhealthy behaviors such as smoking tobacco, eating red meat or drinking alcohol, an d as a result, may help to lower morbidity and mortality rates among certain religious groups (Cochran, Beeghley, & Bock, 1988; Gottlieb & Green, 1994). Regardless of the hypothesis being explor ed, it has been argued that when exploring explanatory mechanisms of relig ious influence on health outcomes, it is necessary to control for the effects of beha vioral traits on outcomes to avoid spurious relationships (Benjamins, 2004). Other resear chers have cautioned that age, ethnicity, gender, education, disability, and poor health should be treated as confounders because of their relationship to both religion or spirituality and health and well-being outcomes (Ainlay, Royce, & Swigert, 1992; Ferraro & Koch, 1994; Le vin, Taylor, & Chatters, 1994). Negative Effects of Religion on Health and Well-being Religion is a complex system of beliefs and teachings that can heavily influence attitudes and behaviors. For example, reli gion may play a role in how people judge and compare themselves, or be a key in defini ng which qualities and be liefs are important. These thoughts and comparisons could either re sult in a view that defines oneself in a positive, self-efficacious context or a nega tive, powerless one (Dull & Skokan, 1995).
52 Such views can affect feelings of empowerme nt, optimism, and self-esteem, (Pargament, et al., 1990), and in turn, influence health and well-being. Religion may reduce feelings of control and create increased stress in individuals who stri ve to conform to religious doctrine and authority (Dull & Skodan). Al so, not all individuals gain a sense of hope and optimism from religious beliefs (Dull & Skokan, 1995). Religious beliefs may play a role in whether people view cu rrent events or challenges with optimism or fatalistic pessimism. Such hopelessness and pessimism can debilitate self-esteem and selfenhancing actions (Peterson, Seligman, & Vaillant, 1988). Causal Considerations and Alternative Explanations The above results should be interprete d with caution (Crowther, Parker, Achenbaum, Larimore, & Koenig, 2002). It has been argued that such findings may not produce conclusive results because religiously committed individuals may be less likely to engage in behaviors that pose threats to health, such as cigarette smoking and excessive alcohol use. They are also fre quently involved in close family systems and supportive communities that may actually account for the causal mechanisms in the religion and health association (Crowther, et al., 2002). An alternative explanation for religionÂ’ s positive effects on well-being might lie in sense of belonging, fellowship and support th at religious involvement often entails. As noted previously, there is ample evidence in the literature that demonstrates the salutary nature of social relation ships (Levin, 1994a, 1994b). El lison (1994) contends that individuals who are embedded in religious communities may enjoy larger and denser social networks than nonreligious individua ls. Church contacts can create opportunities
53 for individuals to become integrated into wi der, more diffuse netw orks that reach beyond the confines of their congregation (McIntos h & Alston, 1982). Churchgoers may receive more social support and may perceive their support networks to be more reliable and more supportive than their non-religiously-ori ented counterparts (Ellison, 1994; Ortega, Crutchfield, & Rushing, 1993). This support may include companionship, provision of a confidant, instrumental aid, informationa l assistance, bereavement support, pastoral counseling, visitation, etc. (Ellison, 1994; Eng, Hatch, & Callan, 1985; Haber, 1984; Ortega, et al., 1993). Measuring Religious Involvement Religious attendance is frequently seen as an indicator of organized religious activity and is one of the most widely used religion measures (Benjamins, 2004). Religious attendance is associated with a wi de range of both physical and mental health outcomes, such as, well-being, self-rated health, hope, optimism, purpose, self-esteem, coping and depression. It is positively associated with a wide range of physical and mental health outcomes, despite the inclus ion of extensive cont rols (Benjamins, 2004; Chatters, 2000; Ellison, 1991; Musick, Tr aphagen, Koenig, & Larson, 2000), and negatively associated with mortality (H ummer, Rogers, Nam, & Ellison, 1999; Strawbridge, Cohen, Shema, & Kaplan, 1997). Religious salience, a measure of priv ate religious involvement, beliefs and attitudes, is associated with a wide range of mental and physical hea lth outcomes such as reduced levels of depression and higher leve ls of well-being (Benjamins, 2004). Salience may be reflected in measures of how fr equently one prays or how important one
54 considers their own religion. For example, adults with high levels of religious commitment, specified as incorporating both religious meaning and religious belonging, report significantly more satisfaction with their lives than persons with lower levels of commitment (Hadaway & Roof, 1978). Additionally, it has been found that the importance attached to religious faith, or reli gious meaning, is a stronger predictor of life satisfaction than subjectsÂ’ marital status or reported number of friends. Religious salience also appears to have a positive effect upon meani ng and purpose in life, even after controlling for age, e ducation, race, marital status perceived health, church attendance and other religiosity variables (Petersen & Roy, 1985). Inconsistency of findings across studies of the relationship between religion and health has been fueled by the absence of adequate measures of religiosity and spirituality (Idler, Musick, Ellison, George, Krause, et al., 2003; Krause, 1993; Levin & Vanderpool, 1987). This inconsistency suggest s that careful attention should be given to the choice of variables used to measure religiosity to reduce spuriousness of findings and enhance generalizability. Multidimensionality of Religious Involvement Social scientists have long recognized the multidimensionality of religiosity, but such conceptual insights have been slow in being implemented in empirical studies with older populations (Krause, 1993). There have been recent advances in the conceptualization of religiosity constructs and their measurem ent (for a review see Hill & Pargament, 2003; Idler, et. al., 2003; Powell, et. al., 2003). Some researchers contend that, in order to tap the complexity of the construct of religiosity, one must address a
55 number of domains that reflect a wide variety of behavioral aspects, such as attendance at worship services, solitary prayer, meditation, re ading of sacred texts, as well as attitudinal aspects such as beliefs, values, and feelings (Glock & Stark, 1965; Krause, 1993; Idler, et. al., 2003). Failure to examine a wide set of dimensions of religiousness and spirituality and when exploring their relations hip to health and well-being may result in vague, inconclusive or spurious results. Mediating Variable Dispositional Optimism There is accumulating evidence that s uggests that dispositional optimism is beneficial for physical and psychological wellbeing (Scheier, Carver, & Bridges, 1994). Dispositional optimism refers to an indi vidualÂ’s perceptions of situations and expectations of favorable or positive outcomes over time (Brenes, Rapp, Rejeski, & Miller, 2002; Scheier & Carver 1985). Conversely, pessim ists are individuals who generally have a more negative outlook on lif e and expect that things will go badly (Scheier & Carver, 1985). Positive emoti ons serve as markers of optimal well-being (Fredrickson, 2001), and optimistic orientations to the future have been identified as important hallmarks of successful aging (Bir ren & Renner, 1980). Previous studies have demonstrated that dispositional optimism is associated with both physical and psychological outcomes, including health symptoms, depressive symptomatology and surgery recovery (Brenes, Rapp, Rejeski, & Miller, 2002; Peterson, 2000; Scheier & Carver, 1992). Pessimism has been associated with anxiety, stress, self-rated health (Robinson-Whelen, Kim, MacCallum, & Kiecolt-Glaser, 1997), and depression (Bromberger & Matthews, 1996). Researcher s have found that persons with a more
56 positive life orientation tend to consider their health better than others, have significantly less clinical depression among th em, are healthier at baselin e, have higher levels of functioning than the rest of the sample, a nd have more favorable outcomes (Pitkala, Laakkonen, Strandberg, & Tilvis, 2004). Individuals who believe that their ac tions will lead to favorable outcomes presumably will continue to put forth effort toward their aims and goals in life, whereas those who believe that failure is inevitable tend to give up their efforts and disengage from the pursuit of their goals (Lucas, Dien er, & Suh, 1996). Optimistic individuals have been shown to adjust more favorably to life transitions than do those persons who hold a more pessimistic life outlook (Aspin wall & Taylor, 1992). Optimists appear to differ from those more pessimistically oriented in that they seem to possess more stable coping tendencies (Carver, Scheier, & We intraub, 1989), display fewer depressive symptoms (Chang, Maydeu-Olivares, & DÂ’Zurilla, 1997), and differ in the manner in which they cope with serious health threats or disease (Carver, Pozo, Harris, Noriega, Scheier, et. al., 1993; Friedman, Nelson, Baer, Lane, Smith, & Dworkin, 1992; Taylor, Kemeny, Aspinwall, Schneider, Rodriguez, & Herbert, 1992). Those with a more pessimistic attitude may seek treatment less often, may be less compliant with treatment regimens, may be less physically active, a nd may utilize less adaptive strategies for coping with pain, all of which might lead to poorer physical outcomes (Brenes, et al., 2002). Conversely, it has been theorized that optimis ts are more likely to participate in healthy behaviors and pursue more health-re lated goals (McNicholas, 2002; Scheier & Carver, 1985).
57 Optimism as an Explanatory Factor in Resilience Outcomes The ability to remain optimistic even in when faced with difficult circumstances, such as the deterioration associated with aging or illness, may represent a reserve resource important to resilient outcomes (Taylor, Kemeny, Aspinwall, Schneider, Rodriguez, & Herbert, 1992; Taylor, Keme ny, Reed, Bower, & Gruenewald, 2000). Positive emotions may actually lead individual s to become more resilient to adversity over time (Fredrickson, 2001). There is ev idence that optimism along with high selfesteem and perceptions of pe rsonal control predic t better post-stressor adjustment and serve as core resources that contribute to a resilient personality (Aspinwall & Taylor, 1992; Carver, Pozo, Harris, Noriega, Scheie r, et. al., 1993; Major, Richards, Cooper, Cozzarelli, & Zubek, 1998). Other research ers have found that optimism is associated with engaging in social re lationships (Jylh & Aro, 1989; Olsen, Olsen, GunnerSvensson, & Waldstrom, 1991), although the direc tion of the associations, are less clear. It is possible that people who are more optim istic or hopeful about their future are more likely to engage in social interactions as well as other activities such as volunteerism, religious activities, and health promotion. Alternatively, social integration may predict optimistic orientation (Symister & Friend, 2003). It is also quite likely that the associations are actually bidi rectional, or are mediated by other factors, such as selfesteem. Regardless of the causal direction of associations, it is posited that optimism or hopefulness will impact respondent behavior and outcomes. Optimism also has been conceptualized in research as positive life orientation. Positive life orientation has been shown to be protective against mortality and permanent institutional care in the elderly (Pitkala, et al., 2004). Persons w ith a positive outlook on
58 life consider their health better than others, ar e less clinically depressed (Ostir, Markides, Black, & Goodwin, 2000; Pitkala, et al., 2004), a nd engage more in social relationships, which have a protective effect against adverse health outc omes (Olsen, Olsen, GunnerSvensson, & Waldstrom, 1991). Control Variables Age (Field & Minkler, 1988; Fung, Carste nsen, & Lang, 2001; Lee & Markides, 1990), gender, education (Feldman, Makuc, Kleinman, & Cornoni-Huntley, 1989; Haan, Kaplan, & Comacho, 1989; House, Kessler & Herzog, 1990; Marmot, Kogevinas, & Elston, 1987; Ulbrich & Warheit, 1989), a nd marital status (Bisschop, et al., 2004; Bosworth & Schaie, 1997; Ross & Mirowsky, 1989) or whether or not the respondent lives alone are associated with resilien ce and successful aging outcomes (Berkman, Seeman, Albert, Blazer, Kahn, et al., 1993; Kahana, et al., 2002; Strawbridge, Cohen, Shema, & Kaplan, 1996). These demographic influences represent relatively immutable influences in a successful aging paradigm. In come or assets, representing financial and environmental resources, are among the most wi dely studied predictors of late-life wellbeing (Larsen, 1978). Income is associated with a number of positive outcomes including successful ag ing (Aneshensel, 2002; Baltes & La ng, 1997). Elderly adults with higher socioeconomic status tend to join more organizations, volunteer more and demonstrate greater social integration, all of which have been associated with successful aging and enhanced wellbeing (Thoits & Hewitt, 2001; Wilson, 2000). Elderly adults with lower income may experience greater e xposure to health thre ats such as unsafe environmental conditions, poor housing, inad equate nutrition (Robert & House, 1996), higher levels of stress and depression, and limited access to resources and assets that
59 might otherwise offset the adverse impact of these stressors (Aneshensel, 1992). Lower income adults also may have greater need of social support, and experience greater adversity when they do not receive adequa te social support (Unger, McAvay, Bruce, Berkman, & Seeman, 1999). Conversely, elderl y adults with higher income are able to afford health care, assistive devices, better nut rition, transportation, and entertainment. Adequate income also allows them the opportunity to pay others for assistance with household tasks (Unger, et al., 1999). The ability to pay for ap propriate housing, household assistance, or travel to visit w ith friends and relatives may enable higher income elderly to compensate for a lack of social support (Unger, et. al., 1999) or substitute for activities they give up as agi ng creates barriers to participation (e. g., religious attendance). Covariates Health Self Appraisal Previous research has shown correla tions between health self-appraisal (subjective health) and both physical and psychological well-being (Bosworth & Schaie, 1997). Many studies have demonstrated that, in older adults, self-ra tings of health are significantly shaped by objective health conditions (e.g., Fillenbaum, 1979; Maddox, 1962; Tissue, 1972), and are frequently simila r to those performed by physicians (LaRue, Bank, Jarvik, and Hetland, 1979; Suchman, Phillips, & Streib, 1958). Global selfevaluations of health have t hus proven to be effective predictors of morbidity and mortality that reflect substantial stability through time (Gold, Fra nks, & Erickson, 1996; Idler & Kasl, 1991; Idler, Kasl, & Lemke, 1990; Kaplan & Camacho, 1983; Maddox &
60 Douglas, 1973; Schoenfeld, Malmrose, Blazer, Gold, & Seeman, 1994; Wolinsky & Johnson, 1992). Although previous works have frequently differed in how health selfappraisals have been measured, the consis tency of effects indi cate the importance of including this construct in studies seeking to predict physical or psychological wellbeing. Positive and Negative Affect Temperament has a strong influence on subj ective well-being. Studies on twins separated at birth have demonstrated that both pleasant and unpleasant affect have a strong genetic basis (Tellegen, Lykken, Bouchard, Wilcox, Segal, & Rich, 1988). Numerous studies demonstrate that negative em otions, such as depression and feelings of loneliness, are significant pred ictors of declining health a nd reduced survival in old age (Fratiglioni, Wang, Ericsson, Maytan, & Winblad, 2000; Kivela, & Pahkala, 2001; Penninx, Guralnik, Ferrucci, Simonsick, D eeg, & Wallage, 1998; Tilvis, Pitkala, Jolkkonen, & Strandberg, 2000). Recent studies demonstrate that mood serves as an important factor in psychologi cal adjustment (Chang & Sanna, 2001). Most importantly, findings indicate that such ps ychological adjustment meas ures as depression and life satisfaction are strongly associated with both negative and positive affect (Chang, Maydeu-Olivares, & DÂ’Zurilla, 1997). Mood is composed of two distinguishable di mensions, identified as positive affect (PA) and negative affect (NA) (Watson, Clar k, & Tellegen, 1988). Positive affect and negative affect are generally considered to represent the emotional components of subjective well-being (Hillers Jorm, Herlitz, & Winblad, 1998). Generally, PA is
61 strongly related to extr oversion (but not neuro ticism) and NA is subs tantially correlated with neuroticism (but not extroversion) (Die ner, 1984; Hillers, et al., 1998). Positive affectivity represents the extent to which individuals generally feel active, alert, and enthusiastic. High PA is characterize d by high energy, full concentration, and pleasurable engagement. Low PA is characterized by sadness and lethargy. Positive affect has been linked to positive physiological changes (Futterman, Kemeny, Shapiro, & Fahey, 1994), better social relationships (Taylor & Brown, 1988), and the practice of more conscientious health habits and appropr iate use of healthcare services (Taylor, Kemeny, Reed, Bower, & Gruenewald, 2000). Po sitive affect is inve rsely correlated with health and environmental factors such as phys ical disability, cognitive problems, negative life events, and minimal social contacts (Pit kala, Laakkonen, Strandberg & Tilvis, 2004). Conversely, negative affectiv ity (NA) reflects a general dimension of subjective distress and unpleasant engagement, and reflects the extent to whic h an individual feels guilt, anger and fear (Hiller s, et al., 1998). Negative emotions, such as depression and feelings of loneliness, are si gnificant predictors of declin ing health status in old age (Kivela & Pahkala, 2001; Pennix, Guralnik, Ferrucci, Simonsick, Deeg, & Wallace, 1998). Low NA is characterized as a state of calmness and serenity (Watson, et al., 1988). Trait PA and NA roughly correspond to the personality factors of extraversion and anxiety/neuroticism, respectively (Wat son & Clark, 1984). Whereas, NA is related to self-reported stress and poor coping (Willis, 1986) and health complaints (Tessler & Mechanic, 1978), PA is related to social ac tivity and satisfaction (Watson, 1988a).
62 In light of the evidence s uggesting the heritabi lity PA and NA, and the possibility that they comprise some proportion of subjec tive well-being, PA and NA were measured at Time 1 and treated as c ovariates for this study. Functional Health / Physical Impairment The negative impact of physical health problems on the quality of life of older adults is well documented in the gerontological literature (Vaillant, 1977). Good health can be considered a resource when dealing in the context of social exchange. Bad health can serve as a constraint (H ogan, Eggebeen, & Clogg, 1993). Functional health predicts survival and successful aging (Laukkanen, Leskinen, Kauppinen, Sakari-Rantala, & Heikkinen, 2000; Scott, Macera, Cornman, & Sh arpe, 1997). Poor health may serve as a barrier to participation in pros ocial behaviors, such as religi ous attendance or other social interactions (Fischer & Schaffer, 1993; W ilson & Musick, 1997a). Age-associated increases in chronic illness and negative self-appraisals of he alth are accompanied by lower levels of morale and a general absen ce of well-being (Bissc hop, Kriegsman, Deeg, Beekman, & van Tilburg, 2004; Larsen, 1978). Researchers have argued that informati on about functional st atus is generally more useful in determining the overall well-b eing of older adults (George & Fillenbaum, 1985). It is likely that functional limitations will have effects on respondentsÂ’ ability to participate in certain behaviors and activitie s, thus creating a selection bias (Hays, Landerman, Blazer, Koenig, Carroll, & Musick, 1998). To determine baseline restrictions on activity due to functional disability, it is appropriate to measure the degree of impairment in those everyday activities that involve functioning in and out of the
63 home, referred to as Activities of Daily Li ving (ADLs) and Instrumental Activities of Daily Living (IADLs) (Lawton & Brody, 1969). IADLs are often considered rough measures of global physical and cognitive func tioning. The IADL scale focuses upon the capacity of the individual to perform tasks necessary to maintain independent living (George & Fillenbaum, 1985). Research dem onstrates that the ADL/IADL scales form a hierarchical relationship that is indicative of a unidimensional construct (Spector, Katz, Murphy, & Felton, 1987). To reduce the possibility of spurious associations related to the influence of underlying health conditions on both religiosity and the out come variables, analytic models were run first without functional health controlled, an d again with those variables controlled. In this way it was possible to better determine the degree to which any apparent benefit of activities is the consequence of spurious functional health factors that may constrain individualsÂ’ activity level in th e first place (Lennartsson and Silverstein, 2001). Research Questions The literature provides a great deal of evidence as to the predictive value of religiosity for markers of resilience and successful aging outcomes in the elderly population. However, there is conflicting ev idence as to the path of influence between these factors Â– direct, indirect, or perhaps me diated or moderated by other factors. There is also a lack of consensus as to what mechanisms may account for the relationships between these variables. For example, the positive associations between religious involvement and resilient outcomes in ag ing adults may be more dependent upon dispositional characteristics or other confounders than upon religious involvement itself.
64 Thus, there are gaps in the current understa nding of how religious involvement exacts its protective influence on aging adults. One purpose of this study was to add to the literature by attempting to further delin eate one possible mechanism through which religiosity casts its protective influence on aging adults. The following research questions served as a foundation for the basis of this investigation: 1. Is there a positive association between religiosity and subjective well-being in elderly adults? 2. Is the association betwee n religiosity and subjective well-being in elderly adults mediated by optimism?
65 Chapter Three Methods Purpose of the Study The purpose of this study is to expl ore the association between religious involvement and subjective well-being in el derly adults. Additionally, the study will explore the explanatory power of optimism as a possible mechanism mediating that association. Identification of these relations hips will help to elucidate the pathways and mechanisms through which religious involve ment influences subjective well-being, which may contribute to re siliency in older adults. Conceptual Framework for the Study The conceptual framework of the st udy is based upon the proposition that religiosity has a direct asso ciation with subjective well-being variables that, in the literature, have been associated with resilience in elderly adults. In addition to the direct effects between religiosity and subjective we ll-being, it was also posited that when optimism is added to the analytic model, the association would be mediated and a greater amount of variance would be explained. A significant result of the mediation model would provide a better unders tanding of one of the mechan isms through which religiosity confers its protective effects in predicting subjective well-bei ng in elderly adults. A better understanding of the mechanisms through which religious involvement asserts its
66 influence may assist those who work with elderly adults in creating more specific interventions or services that are better able to facilitate positive outcomes. Identification of specific mechanisms of influence may facilitate the development of alternative resources that can be expected to produce similar outcomes in the absence of those already known to predict successful aging. Research Hypotheses This study tested the following hypotheses: 1. Religiosity is positively associated with subjective well-being, after controlling for sociodemographic factors, baseline health and function, baseline measures of depressive symptomatology, life satisf action and positive and negative affect. 1.1. Religiosity is positively associated with life satisfaction after controlling for sociodemographic factors, baseline health and function, and baseline measures of depressive symptomatology, life satisf action and positive and negative affect. 1.2. Religiosity is inversely associated with depressive symptomatology after controlling for sociodemographic factors, baseline health and function, and baseline measures of depressive symp tomatology, life satisfaction and positive and negative affect. 2. The association between reli giosity and subjective well-be ing in elderly adults is mediated by optimism, after controlling for sociodemographic factors, baseline health and function, and baseline measures of de pressive symptomatology, life satisfaction and positive and negative affect.
67 2.1. The association between reli giosity and life satisfaction in elderly adults is mediated by optimism, after controlling for sociodemographic factors, baseline health and function, and baseline measur es of depressive symptomatology, life satisfaction and positive and negative affect. 2.2. The inverse association between reli gious involvement and depressive symptomatology in elderly adults is mediated by optimism, after controlling for sociodemographic factors, baseline health and function, and baseline measures of depressive symptomatology, life satisf action and positive and negative affect. Sampling and Data Collection Procedures Data Source There have been a number of cross sec tional and longitudinal studies that have explored the relationships between religion a nd a number of outcomes such as optimism, morbidity and mortality in the elderly. Fewer studies have been conducted utilizing a framework that attempts to identify mechan isms accounting for religiosityÂ’s effects and delineate causal paths between measures of religiosity and selected outcomes commonly associated with successful aging and resilie nce in the elderly. Because many of the previous studies have been confined to cros s-sectional data, they have been limited in their ability to establish causal explanati ons for associations between many of these variables. The Buffers of the Impairment/Disabilit y Cascade Among the Old-Old Study an ongoing, National Institute on Aging-f unded, longitudinal study initiated by the Elderly Care Research Center at Case West ern Reserve University in 1989, provides an excellent opportunity to explor e the influences both religiosity and optimism in an elderly
68 population. The data for the current study were taken from the first, fifth and tenth waves of this NIA study of late-life adapta tion to aging and increasing frailty. This original study, currently in its 16th year of data collection, utilizes a survey design with an extensive interview instrume nt that incorporates both open-ended and closed-ended responses. The questionnaire in cludes questions on socioeconomic factors, self-rated health, chronic illness and disease, pain, fall incidents and concerns, incontinence, functional health, use of h ealthcare and medications, ADLs and IADLs, sensory deficits, depression, self-esteem a nd efficacy, life-style habits, social life, prosocial activity, religiosity, traumatic event hi story, and use of technology. Initial data collection began in 1989 with approximately 1000 adults, 70 or older, who were randomly selected from a large, age-segreg ated, retirement community in Clearwater, Florida. Florida Retirement Study Sample Eligibility criteria for th e original study required that participants be at least 72 years old, reside in Florida at least 9 months of the year be free of major mental and physical infirmities (e.g., confused or bedridden) (Kahana, et al., 2002) and be sufficiently healthy to be able to comple te a 60-90 minute face-to-face interview. All respondents own their one to three bedroom homes within an age-segregated, planned retirement community that is oriented toward independent, activity-based living. There are some support and social / recreational servic es provided within the complex, such as a community-owned bus that transports resi dents to area grocery shopping, exercise equipment, pool, tennis, and shuffleboard available at the cl ubhouse, and minimal
69 volunteer opportunities in the complex (e.g., gift shop, good neighbor, building representatives). The retirement communityÂ’s central location in the broader community affords residents convenient access to a wide range of health care services and social, educational, and leisure and r ecreational resources. The city bus makes stops within the community, and residents can also arrange bus transportation to area attractions, such as the Hard Rock Casino. Respondents in the original study were initially contacted at random using computerized resident listings of the retire ment community. The principal investigators reported a mean age of respondents in the firs t wave of the study as 79.3 years (Kahana & Kahana, 2002). The sample is comprised la rgely of Caucasian residents who have previously migrated from the Midwest (app roximately 50%) or the East coast (30%) (Kahana, Lawrence, Kahana, Kercher, Wisniewski et. al., 2002). Due to its lack of ethnic and cultural diversity, generalizability to the population as a whole may be questioned. However, the sample is consistent with th e population composition of both the retirement community itself and the surrounding community as a whole (U.S. Bureau of the Census, 2000). Consistent with other studies of aging adults, women make up the majority of the sample (67%), a fact that is not surpri sing in as much as they have a greater life expectancy than men (Hooyman & Kiyak, 1999). Original participants in the Florida Retirement study were predominantly of Protestant religious affiliation (68%), and had working class or middle class backgrounds. Respondents were generally we ll-educated, with a mean education level of 14 years (Kahana, et al., 2002). Particip ants are interviewed annually, generally in their home or at an arranged location of the respondentÂ’s choosing (e.g., clubhouse), and
70 are modestly compensated for their time a nd participation ($5.00 gift certificate to a nearby supermarket). Participant Interviews To maximize confidentiality and minimize strain on respondents as they become increasingly frail with age, participants are in terviewed in their homes or, at their request, at the complex community center, by specially trained interviewers The introduction of intra-interviewer bias is redu ced due to the fact that ther e has been minimal turnover in interviewing staff, and the or iginal interviewer for the study has assisted in training newer additions to the staff. The original group of respondents is interviewed each year by a single interviewer who has been with the study since its in ception. Interviews take between 1.5 and 2.0 hours (time tends to increas e with age of responde nt). The interview consists of both closed-ended and open-ended qu estions. Previous research suggests that when personal interviews are used, the use of cards with printe d response choices are helpful (Kutner, Ory, Baker, Schechtman, Hornbrook, Mulrow, 1992). At each interview, response cards are provided fo r each of the close-ended questions. Upon completion of the interview, each respondent is given a copy of the Medical Options Study Short Form (MOS, SF-36) (Ware, & Sher bourne, 1992) and asked to complete the form within the next few days and place it in a self-addressed envelope that has been provided so that it can be mailed directly to the principal investigators for follow-up data collection. Data from annual respondent inte rviews are entered into a central database and stored in SPSS format.
71 The Current Study Three waves of the Florida Retirement Study data were utilized for analysis in this doctoral study, Time 1, Time 5, and Time 10. As is common in longitudinal studies, the original study population has decreased through attrition. The study was initiated in 1989 with 1000 participants. At Time 5, th e database includes 591 respondents and at Time 10, 366 of the original respondents were still participating in annual interviews. It might be argued that given the documented l ack of variance of some variables over time (e.g., optimism, life satisfaction), and the pr obability of moderate change over long periods of time (e.g., 4-9 years) due to th e aging process, the use of three widely separated waves may be less than adequate. However, the characteristics of the dataset not all variables included in this study we re measured at each wave Â– dictated the ultimate choice of waves to be analyzed. For example, optimism is measured at Time 5 and Time 10, but not at Time 1. Time 1 pr ovides descriptive values for the study sample, baseline measures of a number of variables that will be used as controls for the analysis, and measures of the independent variable. Time 5 provides measurements of the proposed mediating variable. Outcome variable s are measured at Time 10. Thus, specific waves of data have been chosen based upon the fact that they offer repeated access to the greatest selection of the vari ables included in this study. Relevant study variables were extracted from each of the three waves chosen for analysis. Data analyses were first complete d on each wave independently. Next, Time 1 and Time 5 data were merged and analyzed fo r those who were in both waves. Finally, data from Time 1, Time 5 and Time 10 were merged and analyzed for those respondents who had valid responses for all three waves of the study.
72 Missing Data The analytic sample included all re spondents with complete data or limited missing data as determined by the following cr iteria: a) One case that had no valid data for any study variables was deleted from anal ysis; b) Cases that were missing responses to entire scales were flagged and moved to a non-sample subset for comparative analyses; c) For variables where Â“refusedÂ” was a re sponse option (e.g., total assets, religious preference), refused was recoded to missing, and the case was moved to the non-sample subset; d) Cases that were missing data fo r stand-alone variables (versus scale items) were moved to the non-sample subset; e) Fo r cases that were missing one or two scale items (except for the independent variable religiosity), mean item scores were created for missing responses; and f) cases missing data for any of the religiosity scale items were moved to the excluded cases subset. These strate gies resulted in a Time 1 sample of 797. Analysis at Time 5 was completed on 532 re spondents, and at Time 10, data was analyzed for 335 respondents. Analyses for merged samples Time 1 and 5 were completed for a sample of 430, and the final sample, comprised of merged Time 1, Time 5 and Time 10 datasets had a final N = 255. Measures Dependent Variables Â– Subjective Well-being Life Satisfaction Moods and emotions, labeled affect, repres ent peopleÂ’s evaluations of the events that occur in their lives (Die ner, et al., 1999). Life satisfaction forms a separate factor from the two major types of affect (PA/NA) (Andrews & Withey, 1976 ; Lucas, Diener, &
73 Suh, 1996). Life satisfaction refers to the conscious cognitively judgmental process that individuals use to assess the quality of thei r lives in accordance with their own unique set of criteria (Shin & Johnson, 1978). Diener and colleagues (1985) argue that, because individuals tend to assign di fferent standards of success to different domains of their lives, it is necessary to assess an individualÂ’s global judgment of his or her life, rather than focusing solely on satisfaction with sp ecific domains. That is the strategy upon which the Satisfaction With Life Scale was developed (SWLS) (Diener, Emmons, Larsen, & Griffin, 1985). The SWLS does not measure all aspects of subjective wellbeing. It is a narrow-band instrument intende d to assess the cognitive rather than affect component of subjective well-being. As su ch, the SWLS items are global rather than specific in nature, and allow res pondents to weight domains of their lives in terms of their own values in determining a global judgment of life satisfaction (Pa vot & Diener, 1993). Normative data for the SWLS are available fo r older adults and a num ber of other diverse populations (see Pavot and Diener, 1993 for listing). The SWLS has been shown to correlate substantially with reports by family and friends of the target subjectÂ’s life satisfaction and with other life satisfaction scales, and has been found to be valid and reliable with elderly respondents (Pavot & Diener, 1993). The SWLS has shown strong internal reliability and moderate temporal stability, with Diener and colleagues (1985) reporting a coefficient alpha of 0.87 for the scale and a 2-month test-retest stability coefficient of 0.82. Similar findings were reported by Pavot, et al (1991) and Yardley and Rice (1991). Evidence for the construct vali dity of the SWLS with other measures of subjective well-being has been demonstrated as well (Pavot & Diener, 1993). The reliability coefficient alpha for the scale in th is study was .82. Initial principal-axis factor
74 analysis indicating a single factor structure accounting for 66% of the variance (Diener, et al., 1985) has been replicated (Pavot & Dien er, 1993; Pavot, Diener, Colvin, & Sandvik, 1991) and shown to be consistent between sexes (Atienza, Balaguer, Garcia-Merita, 2003; Shevlin, Brunsden, & Miles, 1998). Th ere is also strong evidence of construct validity (Pavot & Diener, 1993). The SWLS is presented in figure 2. Strongly Agree Agree Neutral Disagree Strongly Disagree In most ways, my life is close to my ideal 1 2 3 4 5 The conditions of my life are excellent 1 2 3 4 5 I am satisfied with my life 1 2 3 4 5 So far, I have gotten the important things I want in life 1 2 3 4 5 If I could live my life over, I would change almost nothing 1 2 3 4 5 Figure 2 Satisfaction With Life Scale (SWLS) The SWLS was available for each of the three waves analyzed in this study. New variables were created at each wave by su mming the five items for each respondent to represent total life satisfaction scores. There were no missing data for the scale at Time 1 or Time 5. Respondent mean item totals were imputed for three respondents at Time 10
75 that were missing one item score each. Hi gher scores reflect hi gher levels of life satisfaction. No reverse-coding was necessary for this scale since the codebooks for this scale at Time 1 and Time 5 indi cated that response options were flipped at data entry, and that the response items were printed in reverse order on the original interview instrument. The life satisfaction variable at each wave was included in bivariate analyses to determine associations and correlations with other scales and variab les in the study, and was included as a control variable at baseline and as a dependen t variable at Time 10. Depressive Symptomatology One of the most commonly used meas ures of depression and psychological well-being is the Center for Epidemiol ogic Studies of Depression Scale (CES-D) (Radloff, 1977). The original CES-D scale c onsists of 20 items representing depressive symptoms for which participants are asked to indicate the frequenc y of occurrence during the past year. A four-point Likert scale is us ed to score responses ranging from rarely or none (1) to all or most of the time (4). The CES-D has been found to be reliable and valid with older adults (Radloff & Teri, 1986 ; Wallace, et al., 2001), who, except for the severely cognitively demented, have been found to be reliable informants regarding their affective states (Lewinsohn, Seeley, Robert s, & Allen, 1997; Parmelee, Katz, & Lawton, 1989), and has consistently demonstrated reli ability and validity in community surveys (Radloff, 1977). It has been demonstrated that screening tools with high reliability (e.g., reliability of approximately 0.90) can be shortened, sometimes by dropping as many as 75% of the items, without losing much of th eir sensitivity and sp ecificity (Lewinsohn, Seeley, Roberts, & Allen, 1997; Shrout & Yage r, 1989). Revised versions of the CES-D,
76 regardless of scoring methods, have produ ced comparable reliability, sensitivity, specificity, and positive predictive values to those reported for the original 20-item CESD (Boey, 1999; Cheng & Chan, 2005; Irwin, Artin, & Oxman, 1999; Kohout, Berkman, Evans, & Cornoni-Huntley, 1993; Turvey, Wallace, & Herzog, 1999). For this study, levels of depressive symptomatology were meas ured with a revised CE S-D scale that was constructed with ten items from the original CES-D scale that were present in each of the three waves of data. This revised versi on of the CES-D is provided in figure 3 Never/ Rarely Some/ Little Occasionally Most of the time All of the time 1. I felt that I could not shake off the blues, even with help from family or friends 1 2 3 4 5 2. I could not keep my mind on what I was doing 1 2 3 4 5 3. I felt depressed 1 2 3 4 5 4. I was hopeful about the future 1 2 3 4 5 5. I was happy 1 2 3 4 5 6. I felt lonely 1 2 3 4 5 7. I enjoyed life 1 2 3 4 5 8. I had crying spells 1 2 3 4 5 9. I felt sad 1 2 3 4 5 10. I could not Â“get goingÂ” 1 2 3 4 5 Figure 3. Revised CES-D Scale
77 Items 4, 5, and 7 were recoded so that all re sponses were scored in the same direction, with higher scores indicating higher levels of depressive sy mptomatology. To score the scale, in line with similar research, respons e options for the Likert scale were converted to a 0-3 format (0 = never / rarely; 1 = some / little; 2 = occasionally; 3 = most / all the time), and scale totals for depressive symptomatology were computed for each respondent. Scale sum scores could range fr om 0 30. The summed scores were then used to create a new variable at each wave, de p1, dep5 and dep10. The reliability of this revised scale was assessed by calculating coefficient alpha (Cronbach, 1951), and was found to have an internal consistency of 81 (Table 1). Initial baseline scores on this revised CES-D scale are included as a contro l variable in order to reduce chances of confounding. Scores from the revised scale at Time 10 serve as a dependent variable. One hundred seven cases were deleted from the original sample at Time 1 due to having no valid responses for the CES-D items. One case was removed from Time 5 and three cases were removed from Time 10. Five respondents missing one scale item each at Time 10 had their own mean item score inserted for missing scale items.
78 Table 1 Means, Standard Deviations, Intercorrelations, and Coefficient Alpha Reliability Estimates for Time 1 Scales Variables Mean SD 1 2 3 4 5 6 7 8 1. NA1 8.49 3.14 (.80) 2. PA1 16.00 3.80 -.01 (.75) 3. SWLS1 18.73 3.28 -.33 .19 (.82) 4. Dep 1 6.95 4.90 .52 -.30 -.45 (.81) 5. religiosity1 0.01 0.76 .02 .20 .10 -.10 (.87) 6. IADLs 1 1.13 0.41 .16 -. 04 -.10 .20 -.04 (.88) 7. ADLs 1 1.04 0.23 .09 -.04 -.05 .08 -.04 .54 (.84) Note : N = 797. Reliability estimates appear on the diagonal Estimates for Time 5 Scales Variables Mean SD 1 2 3 4 5 6 7 8 1. NA5 9.02 3.78 (.83) 2. PA5 15.06 3.99 -.11 (.78) 3. SWLS5 18.02 2.94 -.31 .32 (.81) 4. Dep5 8.54 5.19 .65 -.44 -.48 (.84) 5. religiosity5 0.00 0.85 .03 .37 .15 -.15 (.91) 6. IADLs 5 1.39 0.56 .22 -. 29 -.25 .41 -.03 (.82) 7. ADLs 5 1.06 0.24 .08 -.13 -.12 .18 -.05 .41 (.79) 8. Optimism 5 21.87 2.63 -.34 .40 .39 -.48 .23 -.17 -.11 (.74) Note : N = 567. Reliability estimates appear on the diagonal Continued on next page
79 Table 1 continued Estimates for Time 10 Scales Variables Mean SD 1 2 3 4 5 6 7 8 1. NA10 9.43 3.90 (.84) 2. PA10 16.46 3.61 -0.15 (.78) 3. SWLS10 18.65 3.65 -.39 .51 (.87) 4. Dep 10 9.72 5.80 .63 -.53 -.64 (.88) 5. religiosity10 .01 0.85 .06 .31 .26 -.07 (.92) 6. IADLs 10 1.55 0.74 .14 -. 04 -.10 .25 -.05 (.91) 7. ADLs 10 1.19 0.45 .06 -.14 -.17 .13 -.02 .74 (.87) 8. Optimism 10 22.44 3.82 -.38 .55 .68 -.68 .22 -.15 -.08 (.86) Note : N = 359. Reliability estimates appear on the diagonal Independent Variable Religious Involvement In line with current literature that sugge sts that religiosity is more adequately measured by a multidimensional scale, six variables were chosen to create a religiosity index for this study. Conceptually, this inde x allows for inclusion of a wide variety of behavioral aspects, such as attendance at worship services, participation in church activities other than formal services, as well as attitudinal and subj ective aspects such as values and beliefs ( I find comfort in religion, I put trust in God, seek GodÂ’s help in stressful times), and salience (e.g., perception of religiousness), all previously shown to be important when exploring the associ ation between religiousness and well-being (Ainlay & Smith, 1984; Glock & Stark, 1965; Krause, 1993; Idler, et. al., 2003). These variables were chosen because they appear in each of the three waves, thus can be used to
80 determine correlations between time points, and because they, or variables very similar to them, have been used in the cu rrent literature, thus facilitating study comparisons. Due to the uniqueness and personal variability of th ese variables, and their importance in formulating the independent variable at ba seline, respondents with missing data on these items at Time 1 were not included in the study sample. Because there were unequal levels of response options for the variables, the scores were standardized prior to the development of a religiosity score, which is based upon the sum of the item responses, with higher total scores represen ting higher levels of religiosity. For descriptive purposes, religious preference is included at Time 1. The Religiosity Index appears in Figure 4. Domain\(Variable) Item Wo rding Response Options Affiliation (RELIGA) What is your religious preference? (1) Protestant (i.e., Methodist, Baptist, Presbyterian.) (2) Catholic (3) Jewish (4) Other Please specify __________ (5) Refused (6) None Continued on next page
81 Domain\(Variable) Item Wo rding Response Options Organizational Religiosity (ATTENDA) How often do you attend religious services? (0) Never (1) Less than once a yr. (2) Once or twice a yr. (3) Several times a yr. (4) About once a mo. (5) 2-3 times a month (6) Several times a wk. (7) Every day (8) Refused (ACT9A) How often do you participate in church activities? (1) Several hrs./day (2) One hr/less /day (3) Several times/week (4) Several times/month (5) Rarely or never Nonorganizational Religiosity (Religious coping) When confronted with stress, I try to find comfort in my religion (1) Never (2) Seldom (3) Sometimes (4)Usually (5)Very Often (Religious coping) When IÂ’m confronted with stress, I put my trust in God (1) Never (2) Seldom (3) Sometimes (4) Usually (5) Very often Continued on next page
82 Domain\(Variable) Item Wo rding Response Options Subjective Religiosity (Salience) How religious do you consider yourself? (1) Not at all (2) Not very religious (3) Somewhat religious (4) Religious (5) Very religious Support / Comfort (Religious support) When I am confronted with stress, I seek GodÂ’s help (1) Never (2) Seldom (3) Sometimes (4) Usually (5) Very often Figure 4. Religiosity Index Psychometric Evaluation of Instruments Many scales included in this study have been used extensively in previous research. For these instruments, established reliability and validity data from the literature have been provided. To be consid ered reliable, an instrument must provide consistent scores upon repeated administrati on, upon administration by alternate forms, etc. (Hatcher, 1994). Thus, reliability reflect s the consistency of the scores that are obtained with the instrument in question. In research that involves the use of questionnaire data, the most commonly us ed method of ascertaining consistency and reliability, are the internal consistency indice s of reliability (Hatcher, 1994). Internal consistency is the extent to which individua l items that are combined in a test or questionnaire correlate with one another or w ith the test total. One of the most widelyused indices of internal consistency reliability in the social sciences is the coefficient alpha (Cronbach, 1951; Hatcher, 1994). Coeffici ent alpha reflects the lowest estimate of
83 reliability that can be expected for an in strument (Hatcher, 1994). A high coefficient alpha indicates that the items that constitute the scale or instrument in question are highly correlated with one another. In most research, coefficient alphas of less than 0.70 are generally seen as inadequate (Nunnally, 1978). Table 1 presents assessments for scale reliabilities for the current study, whic h all demonstrate ac ceptable levels. Since the index used to measure religios ity was created for this study, and thus, no psychometric information is available in the literature, it was necessary to explore the underlying scale structure. Expl oratory factor analysis can be used to determine whether the items included in a scale Â“hang together Â” to tap the same dimension or factor (Pedhazur & Pedhazur Schmelkin, 1991). It is essential for the items to demonstrate adequate correlation with factors in order fo r a total score to be meaningful. Factor analysis then, can be used to help justify the validity of a summated scale used to represent a particular construc t or set of constructs (Munro, 2001). Factor analysis was conducted using the 6 items chosen to create th e religiosity index. The principal factor method was used to extract factors, mini mum eigenvalues of 1.00 were specified, significant factor loadings were set at .40, and promax rotation was specified. The scree plot indicated one general factor, which account ed for 0.97 % of the variance, and factor pattern results, shown in Table 2, indicated that all six items demonstrated significant loadings on the single factor of 0 .41 or grea ter. Rotation could not be completed due to the one-factor solution. Interc orrelations of the variables in the scale are shown in Table 3. Fleury (1998) and Munro (2001) argue that in order for meaningful results to be obtained in a factor analysis, correlations between variables should be substantial and that each variable should be highly correlated wi th at least one other variable. These
84 researchers have noted that, in such circum stances, it is common to look for correlations between 0.30 and 0.70. Each variable in the religiosity index was correlated with at least one other variable at a minimum of 0.34. A subsequent assessment of scale reliability through calculation of coefficient alpha (C ronbach, 1951), demonstrated a moderately high reliability estimate of 0.87 (Table 2). Table 2 Factor Loadings for One-factor Sol ution of the Religiosity Scale Factor Loading Variable Label Description .59 ATTENDA During the past year how often have you attended religious services? .41 ACT9A How often do you participate in church or synagogue activities (not including services)? .71 RELITENA How religious do you consider yourself? .89 COPE5A To what extent do you seek GodÂ’s help during stressful events or illness? .88 COPE13A To what extent do you put your trust in God when confronted with stressful events or illness? .88 COPE21A To what extent do you try to find comfort in your religion when confronted with stressful events or illness?
85 Table 3 Correlations Between Indivi dual Items Comprising the Religiosity Index (N = 797) Variable Attendance Activities How Religious Seek GodÂ’s Help Trust in God Comfort in Religion Attendance 1.00 .57 .47 .43 .44 .49 Activities .57 1.00 .34 .25 .29 .29 How religious .47 .34 1.00 .57 .57 .60 Seek GodÂ’s Help .43 .25 .57 1.00 .84 .80 Trust in God .44 .29 .57 .84 1.00 .81 Comfort in religion .49 .30 .60 .80 .81 1.00 Mediators Optimism Optimistic orientation, the proposed medi ating variable, was measured at Time 5 with a revised version of the Life Orient ation Test (LOT-R) (Scheier & Carver, 1985; Scheier, Carver, & Bridges, 1994). Some inve stigators report that factor analyses of the LOT reveal a two-factor solution, on e consisting of items reflecting positive expectancies, and the other consisting of the negatively worded items representing negative expectancies (Marshall, Wortman, Kusulas, Hervig, & Vickers, 1992; Scheier & Carver, 1985). The original scale develope d by Scheier and colleagues (1994) was meant to assess optimism as a bipolar dimension and contained 10 items, three positively
86 oriented items, four negativel y oriented items and three filler items that are not used in scoring. The test developers recommend th at one of the negatively worded items be eliminated from scoring so that equal numb ers of positively and negatively oriented items are used in calculating a total test score. Correlations between optimism and pessimism measured by the scale have been found to range from very low (-0.02) to moderate (0.57) (Brenes, et al., 2002). Although the developers of the s cale initially treated optimism and pessimism as bipolar opposite s, they subsequently have suggested consideration of evaluating optimism and pessi mism independently (Brenes, et al., 2002; Scheier, Carver, Bridges, 1994). However, th ere is no steadfast rule in the literature precluding the use of the scale to represent on e overall score of optimistic orientation. In more recent research, the LOT has been revised by dropping the filler items to reduce the scale to 6 items Â– three measuring positive optimistic outlook and three measuring more negative outlook. The LOTR has been found to correlate 0.95 with the original version of the Life Orientation Test (Scheier, et al., 1994). An acceptable level of internal consistency for the revised scal e has been demonstrated (CronbachÂ’s Alpha 0.78), test-retest reliability has been shown to be fairly stable over time (0.60 at 12 months; 0.56 at 24 months; 0.79 at 28 m onths), and convergent, discriminant, and construct validity have been supported (S cheier, et. al., 1994; Terrill, Friedman, Gottschalk, & Haaga, 2002). In light of the debate in the literatu re regarding the num ber of constructs represented by the LOT, factor analyses were completed to determine the most parsimonious solution. The best solution appear ed to be a one factor model, with the proportion of variance explained above 100% and five of six items with significant
87 loadings on one factor. A subsequent asse ssment for reliability produced an acceptable coefficient alpha of .73. (Coefficient al pha assessed for optimism and pessimism in a two factor solution produced an unacceptable pessimism scale alpha of .45 and an alpha of .79 for optimism). Data for the proposed mediating variable is taken from the LOT as measured at Time 5. Respondents are read a series of shor t statements and asked to indicate, using a Likert scale, the degree to which they agr ee or disagree with the statements. Three statements are negatively worded and three are positively worded. No filler items are included. Response options were previously flipped in the original dataset. The three negatively worded items are reverse coded. Responses are summed to create a variable representing each respondentÂ’s overall optimis m score with higher scores representing greater optimism. Strongly Agree Agree Neutral Disagree Strongly Disagree In uncertain times, I usually expect the best 1 2 3 4 5 If something can go wrong for me, it will 1 2 3 4 5 I always look on the bright side of things 1 2 3 4 5 IÂ’m always optimistic about my future 1 2 3 4 5 I hardly ever expect things to go my way 1 2 3 4 5 Things never work out the way I want them to 1 2 3 4 5 Figure 5. Life Orientation Test (Optimism)
88 Confounders or Covariates Confounders are variables that may exer t influence on either, or both of the independent and dependent variables. If th e effect of a confounder is not accounted for, the contribution of the causal variable to the outcome is unclear or spurious. In extreme cases, the association between an independe nt and dependent vari able will disappear when the confounding variable is statistically controlled, in dicating that the independent and dependent variables are independent of each other and only covary when the control variable is taken into consideration (Aneshen sel, 2002). Conversely, if the addition of the control variable does alter the relations hip between the independent and dependent variables, or if there is significant left ove r covariation, the asso ciation between them cannot be attributed to spurious relationships (Aneshensel, 2002). Therefore, to reduce the possibility of spurious results, a numb er of control variables suggested in the literature as possible influences on the indepe ndent, mediating, or dependent variables are accounted for in the data analysis for this study. For example, it is anticipated that respondents with greater socio economic resources, such as education and income, or who report higher levels of general and functional health will be more likely to report higher levels of optimism and life satisfaction and lo wer levels of depressive symptomatology. Likewise, it is anticipated that respondents who report higher levels of positive affect and lower levels of negative affect will report hi gher levels of optimism and life satisfaction and lower levels of depressive symptomatology. In addition, investigators (e.g., Ensel & Lin, 1991) have also suggested that age, sex, race, education, fam ily income, and marital status must be controlled when studying depr ession outcomes to avoid spurious results. It also seems reasonable to assume that general health and f unctional ability could
89 influence such variables as religious atte ndance and activity participation. Thus, a number of control variables have been sele cted based upon their previously reported associations or currently presumed relations hips with the study variables. Control variable data is taken from wave one. Physical Health & Functional Ability Physical health and functioning were assessed using the following measures: The chronic conditions subscale of the Olde r Americans Resources Study (OARS) Illness Index (Duke University, 1978) used as an aggregate index of pathology (Fillenbaum, 1978; George & Fillenbaum, 1985); disabili ty, as assessed by the OARS Index of Activities of Daily Living (ADL) and Inst rumental Activities of Daily Living (IADL) scales (Fillenbaum, 1978; George & Fillenbaum, 1985). The OARS Index has been widely used in st udies of health and functioning in late life (George & Fillenbaum, 1985) because of its basic assumption that when assessing the functional level of the elderly, a comprehensive approach must be taken (Fillenbaum, 1988). The OARS has documented psychom etric properties among elderly respondents (Mangen & Peterson, 1982), with some research ers reporting an inter-rater reliability of 0.88 (physical functioning) and 0.74 (mental functioning) (Krach, DeVaney, DeTurk, & Zink, 1996). The physical health index identifi es various physical cond itions and clinical symptoms. Respondents are asked to indicate whethe r they have experienced any of the following chronic conditions over the past year: anemia; arthritis; asthma; glaucoma, cataracts or other eye diseases; emphysema/ch ronic bronchitis; oste oporosis; high blood pressure; heart trouble; diabetes; ulcers; stomach or intestinal disorder s; thyroid, liver or
90 kidney disorders; cancers or leukemia; st roke, ParkinsonÂ’s Dis ease, urinary tract disorders (including prostate); orthopedic problems; skin diso rders. Responses are coded 0 (none) or 1 (respondent reports having the co ndition). The conditions were summed to create a new variable at each wave repr esenting a total OARS Index score. The OARS Index of IADLs and ADLs asse ss respondentsÂ’ degree of performing habitual and universal personal care activ ities. The Physical ADL index focuses on personal maintenance tasks of washing a nd bathing, dressing and putting on shoes, getting to or using the toilet, getting in/out of bed unassisted, eati ng without assistance. Respondents are asked how often, without assi stance, they have difficulty performing the tasks. Possible responses are based on a Li kert scale reflecting th e following responses, (1) never, (2) occasionally, (3) frequently, an d (4) always. To obtain an aggregate score reflecting each respondentÂ’s leve l of disability, scores were first summed then a mean score was tabulated. Higher mean totals indica te greater difficulty. Scores were created for respondents in each wave. Baseline scores were entered as contro l variables. Scores at Time 5 and Time 10 were used for desc riptive and bivariate purposes. CronbachÂ’s alpha reliability estimates for this scale at Time 1 were .83 (Table 1). IADL disability was measured by the total number of the following activities done with difficulty: getting from room to room, going out of doors, walking up and down stairs, doing your own housework, preparing your own meals, shopping for groceries. Possible responses are based on a Likert scale reflecting the following responses, (1) never, (2) occasionally, (3) frequently, and (4) always. Scores were summed then a mean was tabulated to form an aggregate score with higher totals indicating more difficulty/disability. Baseline scores were entered as control variables.
91 Scores at Time 5 and Time 10 were used for descriptive and bivariate purposes. CronbachÂ’s alpha reliability estimates for th is scale at Time 1 were .88 (Table 1). Health Self-Appraisal Health self-appraisal is measured at baseline by respondentsÂ’ responses to the question, Â“In general, do you c onsider yourself to be a very healthy, healthy, fairly healthy, sick, or very sick person?Â” Positive and Negative Affect Recent findings have demonstrated that measures of psychological adjustment, such as depression and life satisfaction, are strongly a ssociated with mood (Chang, Maydeu-Olivares, & DÂ’Zurilla, 1997). Mood has been shown to be composed of two distinguishable dimensions, positive and ne gative affect (Chang, Sanna, Yang, 2003). Positive affectivity reflects the extent to whic h individuals generally feel active, alert, and enthusiastic. Whereas negative affectivity reflects the extent to which individuals generally feel upset or unpl easantly aroused (Watson, et al., 1988). Researchers have demonstrated that the PA and NA dimensions often thought to be opposites, actually emerge as two dominant and relatively i ndependent dimensions of affect (Hiller s, Jorm, Herlitz, & Winblad, 1998; Kercher, 1992; Schumkle, Egloff, & Burns, 2002; Watson & Clark, 1992). Investigators have further s uggested that positive emotions serve as important psychological resources that an in dividual may use in coping with stressful events or in reaction to, or adaptation to, ch anges in life situations, thus facilitating wellbeing (Fredickson & Joiner, 2002; Hobfoll, 1989) Other researchers have suggested that negative affect be controlled in similar analys es because of the biasing effects of common
92 method variance that negative affect may ha ve on self-reports of some of the more emotionally-based variables (e.g., see Symist er & Friend). Chang and colleagues (2003) stress that, when working with a mediation mode l, it is important to demonstrate that the influence of a variable such as optimism on psychological adjustment is not fully mediated by affectivity. Thus, to reduce the pos sibility of spurious effects created by the influence of dispositional characteristics of positive and negative affect on the independent, mediating and depe ndent variables, PA and NA are measured at Time 1 with the PANAS and added to the model as covariates. Construct validity of the PANAS has been reported in Watson et al. (1988). Kercher (1992) demonstrated the viability of using a shortened version of the PANAS with a sample of old-old subjects. The subscales produce reliable and essentially uncorrelated PA/NA subscales that have high discriminant and conve rgent validity and reliability (Kercher, 1992). The psychometric qualities are parallel to those obtained in other studies with younger subjects, and each of the two five-item PANAS subscales demonstrated good factor structure, high discri minant validity and reasonable reliability (Kercher, 1992). The shortened PANAS subscales have been used in this study. Respondents are asked to indicate the extent to which they fe lt ten different emotions over the past year. Sum score variables were created to repres ent PA and NA subscales. The PA subscale included the following adjectives: excited, ent husiastic, alert, inspired, and determined. The NA subscale included: distressed, upset, sc ared, nervous, and afraid. Response items were not at all, a little, somewhat, quite a bit, and very much (coded 1 through 5). Evidence has shown that PANAS measures of PA/NA perform similarly regardless of the
93 time frame used in the assessm ent of mood, such as Â“today, Â” Â“past few weeks,Â” or Â“past yearÂ” (Watson, 1988b). Coefficient alpha for the PA and NA scales in this study were .76 and .79 respectively (Table 1). The revi sed PANAS scale is presented in Figure 6. During the last year, to what extent did you feel the following emotions: Not at all A little Somewhat Quite a bit Very Much a. Excited 1 2 3 4 5 b. Distressed 1 2 3 4 5 c. Upset 1 2 3 4 5 d. Scared 1 2 3 4 5 e. Enthusiastic 1 2 3 4 5 f. Alert 1 2 3 4 5 g. Inspired 1 2 3 4 5 h. Nervous 1 2 3 4 5 i. Determined 1 2 3 4 5 j. Afraid 1 2 3 4 5 Figure 6. PANAS Scale Sociodemographic Variables Although subject race is commonly included as a control variable in social and behavioral research, it was not included in this study due to the fact that in the initial waves of the study, the sample population wa s comprised of only Caucasian subjects.
94Age Age can be considered as controlled by design, due to the fact that possible participants were chosen from a sample population who reside in an age-restricted, retirement community. Age was measured a ccording to the self-reported age in years reported by each respondent in th eir baseline interview. Gender Gender was measured as a dichotomous variable, male or female. Education Educational level was measured as a continuous variable at baseline. Marital Status Dummy variables were created to reflect three possible categories for marital status, never married, married, and previously married. Income/Total Assets To reflect the fact that the vast majority of respondent s for this study are retired, and therefore, not reliant on an occupationa l income, income was operationalized as a self-reported measure of total assets. Respondents were asked if they could give an estimate of the total worth of their familyÂ’s a ssets, including such things as the equity in their home, savings, and investments. Re sponse categories included refused, none, and 14 choices that range from less than $10,000 to $300,000 or more. There were no
95 refused responses for sample cases. Medi an income ($150,000 Â– 174,999) was inserted for 82 respondents with missing data for this variable. Variable Response Options Age Continuous variable Sex 1 = Male 2 = Female Marital Status 0 = Never married 1 = Married 2 = Previously married Years of Formal Education Completed Continuous variable Total Assets 15 levels of income Living Status 1) Alone 2) Not Alone Figure 7. Sociodemographic Control Variab les Selected for Study Inclusion Data Analysis The SAS software program version 9.1.3 (SAS Institute Inc, Cary, NC) was used for data analyses. Data analysis for this study included three aspect s, evaluation of the psychometric properties of measures or inst rumentation where appropriate, descriptive univariate and bivariate analyses, and multivar iate analyses. Description of these steps follows.
96 Univariate Analyses Univariate descriptive anal yses provide appropriate methods for screening the data to ensure that they have been entere d correctly and are being read correctly by the program and for establishing a more in-depth description of the characteristics of the study sample. Frequency distributions of ever y variable in the study were computed to determine the amount and pattern of missing da ta. Further, for cat egorical variables, frequency distributions can be helpful in de termining the need fo r collapsing variable categories or generating dummy variables for si mplification. This procedure was utilized with the marital status variable, reducing the response options to three levels and dummy coding the data accordingly. Means and sta ndard deviations were generated for all continuous variables. These step s are valuable tools for identi fying response values that are out of range or non-meaningful, and in id entifying scale items that require reversecoding. Tests of distribution normality fo r each continuous dependent variable were conducted. Univariate statistics indicated th at the distribu tion of the majority of the continuous variables in the study was genera lly skewed and leptokurtic. Based upon this general characteristic of the data, the Spearman coefficient was used to assess correlations between the ratio, interval and ordinal variables. The Spearman coefficient is a non-parametric, distribution-free test that makes no assumption concerning the shape of the distribution from which the study data ha s been drawn. For this reason, it is often used by researchers when one or both of the vari ables are interval or ratio level, but have skewed or non-normal distributions (Hatcher & Stepanski, 1994).
97 Bivariate Analyses In order to explore the impact of exclusiona ry criteria and loss to attrition, the first task in this process was to compare the characteristics of those respondents who are included in the sample with those that were excluded due to missing data at baseline, and those that were lost to follow-up after Time 1. Descriptive analyses and one-way ANOVAs were used to compare the groups with respect to the vari ables measured at baseline. Bivariate analyses were used to understand the differences between groups in relation to control variables. Significant findings from this analysis were used to determine which variables should be included as controls or covariates in further analyses. Chi square was used to determ ine the association between gender, living arrangement and marital status. ANOVA proce dures were used to explore differences by gender and marital status in levels of other continuous control variab les (age, education, assets) covariates (health self appraisal, baseline depressive symptomatology, life satisfaction, IADLs, ADLs, and chronic conditions), the inde pendent variable (religiosity) and the dependent variables (Tim e 10 depressive symptomatology, Time 10 life satisfaction). Spearman correlation co efficients were calculated to assess the relationships between interval or continuous control variables and covariates (age, education, assets, number of chronic conditi ons, IADLs, ADLs, positive affect, negative affect) and the independent (religiosity) a nd dependent variables (Time 10 depressive symptomatology, Time 10 life satisfaction). Spearman correlations were also tabulated to assess relationships between the independe nt variable (religiosity), the mediating variable (optimism), and the dependent vari ables (Time 10 depressive symptomatology, Time 10 life satisfaction).
98 Analyses to Explore Possible Confou nding by Health and Function Variables It is reasonable to assume that f unctional disabilities or underlying health conditions may confound attempts to identify variables that predic t subjective well-being outcomes such as life satisfaction or depres sive symptomatology. For example, chronic health conditions or physical infirmities could reduce a subjectÂ’s ability to participate in activities or make interactions outside the hom e difficult or impossible. Such constraints could dramatically influence responses to some items included in the religiosity scale (e. g., religious activities participation, religi ous service attendance) and could significantly impact an individualÂ’s outlook on life or appr aisal of health and well-being. Relevant baseline measures were added to the model in an effort to reduce the possibility of spurious associations due to the influence of underlying health conditions or physical disabilities. Further, to explore the magnitude of the impact of chronic health conditions and physical limitations, analyses were in itially completed without controlling for physical health and function. Bivariate anal yses demonstrated significant associations between all three health and function variables and the depe ndent variables that ranged from 0.12 Â– 0.16, with one exception. The a ssociation between baseline ADL scores and Time 10 depressive symptomatology was not si gnificant (0.16; p = .06). In addition, omitting the three health and function variab les from the initial models that included control variables, the remaining covariates and the independent variable reduced the Rsquare for the models. In light of these re sults, and in line with current convention, each of the three health and func tion variables were retained for all future analyses.
99 Control Variables Time 1 Covariates Time 1 Independent Variable Time 1 Proposed Mediator Time 5 Dependent Variables Time 10 Age Gender Marital status Total assets Education Living arrangement Health selfappraisal IADLs ADLs Chronic conditions Positive affect Negative affect Life satisfaction Depressive symptomatology Religiosity Optimism Subjective wellbeing Life satisfaction Depressive symptomatology Figure 8. Variables Selected for Study Inclusion
100 Multivariate Analysis Multiple regression analysis was used to analyze data for the hypotheses in this study. This type of analysis is particul arly appropriate when studying the relationship between naturally occurring predictor and cr iterion variables Â– that is, variables that naturally occur in the real world and ar e not the product of manipulation by the researcher (Hatcher & Stepanski, 1994; OÂ’ Rourke, Hatcher & Stepanski, 2005). Multiple regression analysis is a large sample pr ocedure requiring at least 100 observations, preferably 200, to reduce the possibility of unr eliable results. The greater the number of predictor variables included in the multiple regression equation, the greater the number of subjects needed in order to obtain reliable re sults (Hatcher & Stepanski, 1994). The data used for this study includes an acceptable num ber of observations from which to obtain reliable results. Parameter estimates ( b ), standardized multiple regression coefficients ( ), uniqueness indices and model f it statistics were calculated using ordinary least squares regression. OLS regression is based on the principle that predic ted criterion values should be calculated so that the sum of the squared errors of pred iction is kept to a minimum (Hatcher & Stepanski, 1994). Parameter estimates, or multiple regression coefficients ( b ), are the coefficients that are produ ced when the data analyzed are in raw score form. The b weight for a given predictor repr esent the amount of change in the dependent variable that is associated with a one-unit change in that predictor while holding the remaining predictors in the e quation constant. Sta ndardized regression coefficients ( ), or beta weights, are the regression coefficients that would be obtained if all the variables in the equation were standa rdized to a common metric with the same
101 standard deviation. The uniqueness index for a given predictor is the proportion of variance in the criterion vari able that can be explained by that predictor, beyond the variance that is accounted for by the other predictor variables in the model. The R2 value indicates the percent of vari ance in the dependent variable that is accounted for by the linear combination of predictor variables. A significance test of the R2 tests the null hypothesis that R2 = 0. The test results in an F va lue with an associated p value, representing the probability that an F value as large, or larger th an the one produced by the test would be found if th e null hypothesis were true. Modeling Issues Research is generally driven by the desi re to provide explanation. However, George (1996) argues that re search should not on ly predict outcomes but, should also strive to reveal the proce sses that account for the asso ciations between outcomes and predictor variables. The search for expl anatory mechanisms or causal processes is complex, and often difficult to accomplish. Ofte n, there are direct and indirect mediating as well as moderating effects, and temporal issu es that must be considered in the research model (George, 1996). Similarly, Wallace and colleagues (2001) emphasize that it is necessary to look beyond the basic relati onship between independent and outcome variables and consider the complex relation be tween them, as well as to examine how the two mechanisms combine to produce outcomes. It is this wisdom that drives the analyses model for this study.
102 Mediation Effects Mediation effects are hypot hesized when it is thought that the relationship between the independent and dependent variab les is indirect, or transmitted by means of an intermediary. In this situation, the independent variable influences the mediator variable, which, in turn, influences the outcome variable. Thus, the mediator variable is transmitting the effect of the independent variable to the outcome variable (Aneshensel, 2002). When mediation occurs, the mediat or represents the generative mechanism through which the independent is able to infl uence the dependent variable of interest (Baron & Kenny, 1986). Baron and Kenny (1986) maintain that in order to perform causal steps mediation analyses, four conditions must be met. First, a significant relationship must exist between the independent variable (religiosity) and the dependent variables (life satisfaction; depression). Second, there must be a signi ficant relationship between the independent variable (religiosity) and the mediating va riable (optimism). Third, the mediator (optimism) must be significantly associ ated with the depende nt variables (life satisfaction; depression). And fourth, when the mediator (optimism) is controlled, the previously significant relationship demonstr ated between the independent (religiosity) and dependent variables (life sa tisfaction; depression) must decrease significantly. Thus, the mediator acts to reduce the strength of the relationship between the independent and dependent variables. A mediational model in forms how or why certain effects will occur, and can be tested using a series of regre ssion equations (Baron & Kenny, 1986). In a mediational model, it is anticipated that th e relationship between the outcome variables and the independent variables will be redu ced once both the indepe ndent variable of
103 interest and the mediator are in the e quation (Wallace, et. al., 2001). Complete mediation would be indicated if the associ ation between religios ity and optimism and between optimism and subjective well-bei ng (depression, life satisfaction) were significant, but previously si gnificant associations between religiosity and subjective well-being became nonsignificant after controllin g for optimism (James & Brett, 1984). Partial mediation is demonstrated when the re gression coefficient is greatly reduced when the mediator is controlled (Lin, 1997 ), but still remains significant. Although Baron and KennyÂ’s (1986) recommen dations have been widely cited, they come under criticism from other resear chers. Shrout and Bolger (2002) discuss limitations of their approach, suggesting, for example, issue might be taken with the use of a partial mediation model approach when a model of complete mediation has been hypothesized. Baron and Kenny have argued, howe ver, that models in which there is only partial mediation are more realistic in mo st social science research since a single mediator cannot be expected to complete ly explain the relationship between an independent and dependent variable. Nevert heless, many researchers (e. g., MacKinnon & Dwyer, 1993; MacKinnon, Lockwood, Hoffm an, West, & Sheets, 2002; MacKinnon, Warsi, & Dwyer, 1995; Preacher & Hayes, 2004) suggest that there are more formal and statistically rigorous methods by which me diation hypotheses may be assessed. For example, a procedure developed by Sobel (1982 ) may be used to provide a more direct test of a significant mediati on effect. The Sobel test, a produc t of coefficients procedure, estimates and tests the intervening variable effect by dividing the estimate of the intervening variable effect, by its standard error and comparing this value to a standard normal distribution (MacKinnon, et al., 2002). A calculation tool for obtaining
104 Sobel statistics, drawn from MacKinnon & Dwyer (1993) and MacKinno n, et al. (1995) is provided by Kris Preacher at h ttp://www.unc.edu/~pr eacher/sobel/sobel.htm A z statistic may be calculated after entering the alpha, beta, and their standard errors as obtained in regression analyses, or by providing t statistics for alpha and beta. An absolute value of greater than 1.96, assuming a 2-tailed test of 0.05, is considered significant. Overview of Regression Model Building Process A number of steps were used to build the regression models for religiosity and each dependent variable. Initially, sociodem ographic control variables were entered as a block to account for as much variance in the de pendent variables as possible. At Step 2, the remaining proposed confounders (chronic conditions, baseline depression score, ADL and IADL composite scores, PA/NA score, life satisfaction score) were added one by one to determine the amount of variance accounted fo r in the overall model. For Step 3, the predictor variable (religiosity) was added to determine its relationship to the dependent variables (depressive symptomatology, life satisfaction) For Step 4, the proposed mediator (optimism) was entered in place of the dependent variables in order to determine its relationship with the independent variable (religiosity). Finally, the 5th Step was to determine if the effect of the independent variable (religiosity) on the dependent variables (Time 10 depressive sy mptomatology, Time 10 life satisfaction) would be reduced when the mediator (optimis m) was included in the model. When all variables of interest had been entered into the regression model, a series of diagnostic tests were conducted to look for outliers, in fluence, leverage, and collinearity.
105 In interpreting results from OLS regr ession procedures, the model R-square indicates the percentage of variance in the de pendent variable that is accounted for by the linear combination of the control and indepe ndent variables. The model R-square was used to assess the overall ability of the model containing independent and control variables to predict each dependent variable. The significance of each of the predictor variables, after controlling for all other variables in the models, in predicting outcome variables was assessed by the p-value asso ciated with their standardized regression coefficients (beta weights). In multiple regression procedures one is interested not only in the amount of variance accounted for, but also in the statis tical significance of each of the independent variables. The fact that the R2 is statistically significant doe s necessarily indicate that all the independent variables are contributing si gnificantly to the explained variance. Testing the beta weights for each independent variable will indicate whether the variable is contributing significantly to the variance acco unted for in the dependent variable. Beta weights are the regression coefficients that would be obtained if all the variables were standardized and had the same standard devia tions. A p-value of less than 0.05 was used as the criterion for assessing statistical signi ficance and rejecting th e null hypothesis that the regression coefficient is equal to zero. Lastly, the uniqueness index was determined for each of the predictor variables and an F-test was used to test for statistical significance to determine if the nu ll hypothesis was equal to zero.
106 Design Limitations Due to the fact that this study utilized secondary data analysis, the selection of variables representing the constructs of in terest is governed by available data. For example, the variables included in successi ve waves are not always identical, thus restricting the ability to util ize variables based upon passage of time. Additionally, in some cases, it was necessary to forego multidimensional measurement of constructs due to lack of measures included in the original study instrume nts. For example, although recent literature suggests the necessity of operationalizing religious involvement or religiosity as a multidimensional construct encompassing as many as six dimensions, in early waves of the original study, fewer dimens ions are addressed. This will make direct comparisons across similar studies somewhat difficult. To maintain statistical power, it was n ecessary to limit the number of variables considered for analyses. A larger samp le size may have allowed the addition of variables to the analytic model. For exampl e, constructs frequently intertwined with religious involvement, such as volunteeris m and social support, may provide further explanatory power to a similar investigation. Likewise, it might be helpful to include as antecedent factors dispositional variables su ch as altruistic orientation and other personality traits. Power issues also lim ited the ability to operationalize the dependent variable (SWB) as a more inclusive multidimensional measure. For example, it would have been preferable to include additional measures of life satisf action, and variables related to social comparison, goals, mean ing and purpose in life, and respondentÂ’s perception of the quality of their life.
107 Chapter Four Study Sample A total of 255 of the orig inal 1000 respondents met the in clusion criteria for this study, which included: (1) Participation at Ti me 1, Time 5 and Time 10; (2) Complete data for baseline measures of sociodemographic characteristics, health self-appraisal, and chronic health conditions and religiosity measures; (3) Complete data for the proposed mediating variable, Time 5 optimism; (4 ) No more than two missing items on the remainder of the scales used in the study Â– revised CES-D (depressive symptomatology), SWLS (life satisfaction), (PANAS) positive an d negative affect, IADL and ADL scales (functional health). Differe nces between respondents in the final study sample and respondents who were either lost to attrition or were excluded due to missing data were explored with frequency dist ributions, chi-square tests of independence and one-way ANOVAs. Characteristics of the Samp le Population at Baseline Data analyses at Time 1 were co mpleted on 797 of the original 1000 study respondents. The mean age of this sample of elderly adults was 79. Women ranged in age from 71 to 95 years of age, with a mean of 78. Men ranged in age from 72 to 93 years of age, with a mean of 79. As coul d be expected, given the trend for women to have higher life expectancies than men (Hooyman & Kiyak, 1999), female respondents (N=503) outnumbered male respondents (N=294) in the study sample. All respondents in
108 the initial wave of the orig inal study were Caucasian. In line with current national demographics, men (76%) were twice as likely as women (38%) to be currently married, and women were considerably more likely to report being previously married (59%) (e.g., divorced; or most frequently, widowed) than were the male respondents (21%), and over twice as likely as males (26%) to be living alone (59%). The sample for this study was relatively well educated, with a mean edu cational level of almost 14 years (SD 2.54). Women reported slightly less education (mean = 13 years) than men (mean=14 years). The median level of financial assets for the sample ranged from $150,000 $175,000, with women reporting mean levels below the median ($125,000) and males reporting mean levels above the median level ($175,000$199,000) of total financial assets. Table 4 summarizes comparisons between respondents included in initial data analyses (N = 797), respondents who were excluded due to missing data (N = 202), respondents lost to attrition (N = 536), and respondents included in the final study sample (N = 255). Table 5 explores possible gender differences in the measures used in the study.
109 Table 4 Comparison of the Characteristics of Study Participants with Respondents Who Were Excluded Due to Missing Data or Lost to Attrition Characteristic Sample Population at Time 1 ( N = 797) Mean (SD) or n (%) Excluded Due to Missing Data ( N = 202) Mean (SD) or n (%) Cases Lost to Attrition ( N = 536) Mean (SD) or n (%) Final Sample ( N = 255) Mean (SD) or n (%) Age <75 75-79 80-84 85+ 78.6 (4.17) 117 (14.7) 365 (45.1) 250 (31.4) 65 ( 8.6) 82.2 (5.12) 18 (9.0) 53 (26.4) 62 (31.0) 69 (38.8) 79.3 (4.34) 61 (11.4) 221 (41.2) 195 (36.4) 59 (11.0) 77.2 (3.42) 55 (21.6) 142 (55.7) 52 (20.4) 6 (2.4) Gender Male Female 294 (36.9) 503 (63.1) 48 (23.8) 154 (76.2) 205 (38.2) 331 (62.0) 89 (34.9) 166 (65.1) Race (White) 797 (100.0)202 (100.0)536 (100.0) 255 (100.0) Education <12 years 12 years >12 years 13.6 (2.54) 93 (11.7) 218 (27.3) 486 (61.0) 13.4 (2.70) 36 (12.8) 59 (29.2) 107 (45.5) 13.5 (2.58) 70 (13.1) 139 (26.0) 327 (60.0) 13.7 (3.42) 23 (9.0) 77 (30.2) 155 (60.8) Marital Status Never Married Married Previously Married 18 (2.3) 417 (52.3) 362 (45.4) 23 (11.4) 55 (27.2) 124 (61.4) 9 (1.7) 274 (51.1) 253 (47.2) 8 (3.1) 139 (54.5) 108 (42.3) Continued on next page
110 Table 4 (Continued) Comparison of Study Participants with Re spondents Who Were Excl uded Due to Missing Data or Who Were Lost to Attrition Characteristic Sample Population at Time 1 ( N = 797) Mean (SD) or n (%) Excluded Due to Missing Data ( N = 202) Mean (SD) or n (%) Cases Lost to Attrition ( N = 536) Mean (SD) or n (%) Final Sample ( N = 255) Mean (SD) or n (%) Living Arrangements Living Alone Not Living Alone 371 (46.5) 426 (53.4) 142 (70.3) 60 (29.7) 254 (47.4) 282 (53.0) 115 (45.1) 140 (54.9) Total Assets (Median) $150K $175K$100K $124K $150K $170K $150K $175K Health Self Appraisal (1) Very Sick (2) Sick (3) Fairly Healthy (4) Healthy (5) Very Healthy 3.96 (0.82) 2 (0.2) 13 (1.6) 232 (29.1) 320 (40.1) 230 (28.9) 3.96 (0.91) 0 12 (5.9) 51 (25.2) 72 (35.6) 67 (33.2) 3.84 (0.83) 2 (0.4) 12 (2.2) 183 (34.1) 209 (39.0) 130 (24.2) 4.20 (0.74) 0 1 (0.4) 48 (18.8) 108 (42.3) 98 (38.4) Chronic Conditions 0 1 2 3 4 5 6 7 8 9 2.33 (1.62) 96 (12.0) 180 (22.5) 183 (23.0) 164 (21.0) 85 (10.7) 60 (7.5) 18 (2.3) 10 (1.2) 0 0 2.50 (1.88) 26 (12.9) 41 (20.3) 46 (22.8) 36 (17.8) 29 (14.4) 9 (4.5) 8 (4.0) 3 (1.5) 1 (0.5) 3 (1.5) 2.37 (1.63) 61 (11.4) 123 (23.0) 119 (22.2) 113 (21.1) 58 (10.8) 40 (7.5) 14 (2.6) 7 (1.3) 1 (0.2) 0 1.96 (1.53) 44 (17.2) 71 (28.0) 58 (23.0) 37 (14.5) 31 (12.2) 9 (3.5) 2 (0.8) 3 (1.2) 0 0 Continued on next page
111 Table 4 (Continued) Comparison of Study Participants with Re spondents Who Were Excl uded Due to Missing Data or Who Were Lost to Attrition Characteristic Sample Population at Time 1 ( N = 797) Mean (SD) or n (%) Excluded Due to Missing Data ( N = 202) Mean (SD) or n (%) Cases Lost to Attrition ( N = 536) Mean (SD) or n (%) Final Sample ( N = 255) Mean (SD) or n (%) Baseline ADLs (Mean item score) 1.04 (0.23)1.08 (0.28)1.06 (0.27) 1.02 (0.12) Baseline IADLs (Mean item score) 1.13 (0.41)1.29 (0.59)1.17 (0.47) 1.07 (0.25) Baseline Affect Scores Positive Affect Negative Affect 16.00 (3.80) 8.49 (3.14) 14.89 (3.78) 8.03 (3.52) 15.77 (3.82) 8.49 (3.12) 16.54 (3.69) 8.47 (3.18) Baseline CESD 6.95 (4.91)6.71 (6.67)7.19 (4.93) 6.46 (4.85) Baseline Life Satisfaction 18.73 (3.28)17.51 (3.74)18.56 (3.22) 19.11 (3.39) Religious Preference Protestant Catholic Jewish Other None 532 (66.7) 177 (22.2) 49 (6.1) 11 (1.3) 28 (3.5) 149 (73.8) 27 (13.4) 7 (3.5) 6 (3.0) 13 (6.4) 358 (66.8) 125 (23.3) 27 (5.0) 5 (1.0) 21 (3.92) 170 (66.7) 51 (20.0) 22 (8.6) 5 (2.0) 7 (2.7) Continued on next page
112 Table 4 (Continued) Comparison of Study Participants with Res pondents Who Were Excl uded Due to Missing Data or Who Were Lost to Attrition Characteristic Sample Population at Time 1 ( N = 797) Mean (SD) or n (%) Excluded Due to Missing Data ( N = 202) Mean (SD) or n (%) Cases Lost to Attrition ( N = 536) Mean (SD) or n (%) Final Sample ( N = 255) Mean (SD) or n (%) Religious Attendance 0) Never 1) < 1 time per year 2) 1 -2 times per year 3) Several times per year 4) 1 time per month 5) 2-3 times per month 6) Every week 7)Several times per week 8) Every day Missing 4.2 (2.42) 108 (13.5) 51 (6.4) 66 (8.7) 67 (8.4) 30 (3.7) 55 (6.9) 347 (43.5) 62 (7.7) 11 (1.3) 0 3.1 (2.75) 69 (34.3 10 (5.0) 18 (9.0) 10 (5.0) 6 (3.0) 8 (4.0) 69 (34.3) 10 (5.0) 1 (0.5) 1 4.0 (2.45) 84 (15.8) 41 (7.7) 46 (8.6) 51 (9.6) 22 (4.1) 28 (5.2) 228 (42.8) 33 (6.2) 0 3 4.7 (2.21) 24 (9.4) 10 (3.9) 18 (7.1) 15 (5.9) 8 (3.1) 8 (3.1) 27 (10.6) 116 (45.5) 29 (11.4) 0 How Religious 1) Not at all 2) Not very 3) Somewhat 4) Religious 5)Very Religious Missing 3.6 (0.98) 29 (3.6) 73 (9.1) 224 (28.1) 345 (43.2) 126 (15.8) 0 3.5 (1.04) 12 (6.1) 22 (11.1) 47 (23.7) 93 (47.0) 24 (12.1) 4 3.5 (1.00) 18 (3.4) 54 (10.1) 151 (28.2) 231 (43.1) 82 (15.3) 0 3.6 (1.00) 11 (4.3) 19 (7.4) 71 (27.8) 110 (43.1) 44 (17.2) 0
113 Table 5 Gender Differences in Descriptive Characte ristics of Sample at Baseline (N = 797) ____________________________________________________________________________________________________________ Women Men (N = 503) (N = 294) ____________________________________________________________________________________________________________ Sociodemographic Mean (SD) Mean (SD) Chi Sq (df) Characteristic or n (%) or n (%) Age (yrs) 78.31 (3.99) 79.12 (4.42) Marital status 109.79 (2) ** Never married 11 (2.1) 7 (2.3) Married 193 (38.3) 224 (76.1) Previously married 299 (59.4) 63 (21.4) Living arrangements 82.88 (1) ** Alone 296 (58.8) 75 (25.5) One -Way ANOVA Results F Education (years) 13.3 (2.29) 14.0 (2.88) Total financial assets $125 Â– 150K $175-199K ____________________________________________________________________________________________________________ Physical and Functional Mean (SD) Mean (SD) F Health ____________________________________________________________________________________________________________ Self Appraisal of Health 3.96 (0.84) 3.95 (0.77) 0.05 Chronic conditions 2.43 (1 .65) 2.18 (0.77) 4.67* IADLs 1.17 (0.45) 1.08 (0.32) 8.12* ADLs 1.05 (0.27) 1.02 (0.15) 3.76 ____________________________________________________________________________________________________________ Psychological Wellbeing Mean (SD) Mean (SD) F Characteristics ____________________________________________________________________________________________________________ Negative Affect (NA) 8.98 (3.38) 7.66 (2.45) 34.03* Positive Affect (PA) 16.34 (3.71) 15.41 (3.88) 11.19* Satisfaction w/ Life (SWLS) 18.45 (3.49) 19.22 (2.83) 10.17* Depressive symptomatology 7.62 (5.16) 5.82 (4.20) 25.69* ____________________________________________________________________________________________________________ Women Men (N = 503) (N = 294) ____________________________________________________________________________________________________________ Religiosity Mean (SD) Mean (SD) F Variables ____________________________________________________________________________________________________________ How Religious 3.71 (0.95) 3.37 (0.99) 25.53* Religious Attendance 4.40 (2.41) 3.97 (2.40) 5.93* Religious Activities 1.51 (0 .79) 1.42 (0.74) 2.69 Seek GodÂ’s Help 4.17 (1.14) 3.55 (1.37) 47.05* Trust in God 4.21 (1.09) 3.75 (1.27) 29.11* Find Comfort in Religion 3.95 (1.22) 3.33 (1.40) 42.68* Indicates statistical significance at p .05. ** Indicates statistical significance at p<.001
114 Physical Health and Function As can be seen in Table 4, the study population was generally healthy. With regard to physical health a nd function, respondents reported moderate levels of selfappraised health, with an overall mean score fo r the sample of 3.96 on a five-point scale. Over 69% of the sample indicated that they considered themselves healthy to very healthy. A list of 20 physical conditions fr om the OARS index of chronic conditions was used to ascertain the level of chronic dis ease or number of physical ailments each respondent was burdened with at baseline. The mean number of chronic conditions reported was 2.33 (SD 1.62), with women repor ting slightly higher numbers of conditions (mean = 2.43) than males (mean = 2.18). Arthritis was the most common condition reported, with almost 54 % of the sample indicating that they were living with the condition, followed by approximately 33 % repo rting high blood pressure. The sample reported relatively low levels of functional disability, as evidenced by low mean item totals for both Instrumental Activities of Da ily Living (IADLs) (M =1.13, SD = 0.41) and Activities of Daily Living (ADLs) (M = 1.04, SD = 0.23). Item scores for each of the scales ranged from 1 Â– 4, with higher scores i ndicating greater levels of disability. Mean levels of physical disability were similar for male and female respondents. Psychological Well-being Table 4 includes a summary of the measur es of psychological well-being utilized in the study. At baseline, study participants reported moderate mean levels of positive affect (16.0) and satisfacti on with life (18.7), both of which had total response ranges between 5 and 25. Women reported higher leve ls of positive affect (PA) (16.3) than
115 males (15.4), but lower levels (18.4) of life satisfaction in comparison to males (19.2). Mean scores for negative affect (NA) for the sample were 8.4, with women reporting higher levels (8.9) as compared to men (7.6 ). Depressive symptomatology was measured with a revised, shortened version of the CESD. As previously explained, once response categories were recoded and reversed as appr opriate, a sum score was created to form the measure of depressive symptomatology in the current study. Overall, the study population reported a mean level of depressive symptomatology of 6.9 at baseline, with women (7.6) reporting somewhat higher mean scores than men (5.8). Values for depressive symptomatology ranged from 0.00 Â– 27.00 on a scale with a total possible score of 33. Religiosity As Table 4 shows, a majority of th e respondents identified themselves as Protestant (67%). Just over 22% identified their religious preference as Catholic. Of the remaining participants, 6% reported their re ligious preference as Jewish while 4% indicated no preference. Respondents were aske d to assess how religious they considered themselves. Using a Likert scale to measure their response, with higher scores indicating higher self-assessment of religi osity, over half of the sample (59%) indicated that they regard themselves as religious or very reli gious. Women (19%) were twice as likely as men (10%) to identify themselves as very religious (see Table 5). A fairly large proportion (28%) regarded themselves to be somewhat religious. In contrast, a very small proportion (4%) indi cated that they were either no t at all religious or not very religious (9%). Regarding attend ance at religious services ov er the past year, respondents
116 most commonly reported weekly attendance (43% ), with women appearing more likely to attend weekly (46%) than men (40%). More infrequent attendance (e.g., 1 Â– 12 times per year) was reported by slightly more than 34% of the sample. More frequent attendance was reported by less than 10% of the partic ipants. The remaining subjects (13%) indicated that they had not attended reli gious services in the previous year. Subjects Excluded Due to Missing Data at Baseline Respondents who did not have complete data at baseline were flagged during initial analyses for comparison with those who did have complete data as well as the final study sample and the group of respondents w ho were lost to at trition. Baseline characteristics of excluded respondents at baseline are summarized and compared with the other groups in Table 4. The group of respondents excluded due to missing data had a mean age of 82.2 years, higher than bot h the study sample and the group lost to attrition. Over three quarte rs of the group was female (76%), a majority of whom were previously married (61%) and living alone. Respondents in this sample reported similar educational levels to the other samples, but lower total assets than those respondents in the other groups. Of all the groups compared for analyses, they reported the greatest number of chronic conditions, highest mean leve ls of functional difficulties, lowest levels of positive affect and lowest le vels of life satisfaction. Th e excluded group also reported the lowest levels of religiosity among the groups who were compared. One way ANOVAs were computed to de termine if there were significant differences between the study sample and th e excluded sample on baseline measures of the covariates and the independe nt variable. As Table 6 indicates, there were significant differences between the two groups on only th ree of the measures. Those respondents
117 excluded from analysis at baseline tended to re port significantly lowe r levels of positive affect and life satisfaction, and higher leve ls of IADL difficulties. There was no significant evidence that those who were excl uded from the study due to missing data had higher levels of depressive symptomatology, a more negative appraisal of their health, greater negative affect, or more chronic illn ess than those who provided more complete data at the baseline interview.
118 Table 6 ANOVA Results Summarizing Group Differences in Measures Be tween Included Sample Population and Cases Excluded Due to Missing Data and Between Study Participants an d Respondents Who Were Lost to Attrition Variable Time 1 Sample Population ( N = 797) Mean (SD) Time 1 Excluded Cases ( N = 202) Mean (SD) Model F Cases Lost to Attrition ( N = 536) Mean (SD) Final Sample ( N = 255) Mean (SD) Model F Health Self Appraisal 3.96 (0.82) 3.96 (0.91) 0.00 3.84 (0.83) 4.20 (0.74) 31.65* Chronic Conditions 2.33 (1.62) 2.50 (1.88) 1.58 2.37 (1.63) 1.96 (1.53) 11.54* Baseline ADLs (Mean item score) 1.04 (0.23) 1.08 (0.28) 3.12 1.06 (0.27) 1.02 (0.12) 4.82* Baseline IADLs (Mean item score) 1.13 (0.41) 1.29 (0.59) 18.91* 1.17 (0.47) 1.07 (0.25) 9.78* Baseline CESD 6.95 (4.91) 6.71 (6.67) 0.02 7.19 (4.93) 6.46 (4.85) 3.80 Baseline Positive Affect 16.00 (3.80) 14.89 (3.78) 13.76* 15.77 (3 .82) 16.54 (3 .69) 7.22* Baseline Negative Affect 8.49 (3.13) 8.03 (3.52) 3.24 8.49 (3.12) 8.47 (3.18) 0.01 Baseline Life Satisfaction 18.73 (3.28) 17.51 (3.74) 21.17* 18.56 (3.22) 19.11 (3.39) 4.84* Continued on next page
119 Table 6 (Continued) ANOVA Results Summarizing Group Differences in Measures Be tween Included Sample Population and Cases Excluded Due to Missing Data and Between Study Participants an d Respondents Who Were Lost to Attrition Variable Time 1 Sample Population ( N = 797) Mean (SD) Time 1 Excluded Cases ( N = 202) Mean (SD) Model F Cases Lost to Attrition ( N = 536) Mean (SD) Final Sample ( N = 255) Mean (SD) Model F Religiosity Measures Religiosity 3.50 (1.04) 3.24 (1.15) 9.90* 3.42 (1.04) 3.64 (1.01) 7.51* How Religious 1.56 (1.10) 1.61 (1.32) 0.28 3.57 (0.98) 3.62 (0.97) 0.39 Attendance 4.24 (2.42) 3.14 (2.75) 31.40* 3.96 (2.45) 4.68 (2.21) 15.53* Religious Activities 1.48 (0.77) 1.31 (0. 63) 8.71* 1.43 (0.73) 1.58 (0.85) 6.66* Seek GodÂ’s Help 3.95 (1.27) 3.84 (1.38) 0.97 3.90 (1.28) 4.03 (1.23) 1.76 Trust in God 4.04 (1.18) 3.97 (1.31) 0.50 4.01 (1.19) 4.10 (1.74) 0.85 Comfort in Religion 3.72 (1.32) 3.71 (1.38) 0.02 3.66 (1.35) 3.85 (1.26 3.43 Indicates statistical significance at p .05.
120 Table 6 includes results of one way ANOVAs computed to explore differences between the study sample and excluded samp le on religiosity measures. Analyses indicated that there were signi ficant differences in mean le vels of overall religiosity, religious attendance and participation in re ligious activities. Respondents excluded due to missing data at baseline reported significantly lower levels of religiosity, attendance at religious services and participation in re ligious activities. However, there were no significant differences between the groups in self appraisal of religiousness, nor in respondentsÂ’ reported ability to trust in God, find comfort in religion or their tendency to seek GodÂ’s help in stressful situations. Exploring the Implications of Attrition As noted earlier, the original study bega n with 1000 adults at baseline. Not surprisingly, the study population was significan tly effected by attrition. The number of participants completing interviews at Time 5 fell to 591, and only 366 respondents completed interviews at Time 10. In order to be included in the final sample for the current study, respondents were required to have complete (or minimally missing) data for all time points. To expl ore potential implications of participant attrition on study outcomes, baseline characteristics of responde nts who were interviewed at each of the three time points were compared with those of respondents who had complete data at baseline but were missing data for one or both of the subsequent time points. This group of respondents was considered to be lost to attrition. The group of respondents who were lost to attrition (N=703) had a mean age of 80.2 years. Over one half of the sample was female (65%), previously married (51%) and liv ed alone at the baseline interview. They were similar to the study population in relati on to median level of assets ($150K-175K),
121 and mean number of chronic conditions (2.38) Table 4 summarizes group characteristics for respondents lost to attrition and compares them to those respondents included in the final study sample and those excluded due to excessive missing data at baseline. One way ANOVAs were computed to determine whether the group lost to attrition differed significantly from the fina l study sample on measures of health selfimage, positive and negative affect, life satis faction, depressive symptomatology, or functional ability. With the exception of negative affect and depressive symptomatology, significant differences between the groups we re found for each of these variables (all F s >1, p .05). Respondents lost to at trition were significantly mo re likely to report lower levels of health self-image, positive affect and life satisfaction, and higher levels of functional disability. But there was no signi ficant evidence that re spondents lost to attrition had greater levels of depressive symptomatology or a more negative outlook on life. The ANOVA summary results are presented in Table 6. There were also significan t differences between the re spondents lost to attrition and those included in the final study sample related to the religiosity measures. in Table 4 it can be seen that, compared to the final study sample, the respondents lost to attrition reported significantly lower levels of overall religiosity, religious attendance, religious activity, ability to find comfort in religion, and were less likely to seek GodÂ’s help in times of stress. They did not differ signi ficantly, however, in how religious they considered themselves or in thei r ability to trust in God.
122 Comparing the Study Sample to th e Elderly Population of Florida Examination of census data (U. S. Department of Health and Human Services, 2005) indicates that the study popul ation for this research differs in many ways from the elderly population of Florida in general. Many of the differences can be attributed to the original study inclusion criteria that require d subjects to be reside nts of a middle-to-upper socioeconomic class, deed-restricted retireme nt community, to be in good health and to be free of mental and physical infirmities th at would prohibit participation in a lengthy interview. However, principle investigators noted at the outset th at their study sample did reflect both the general community from which it was drawn and the elderly retirement population of Florida in general. Table 7 provides a summary of selected differences between the study sample and th e general population of White, non-Hispanic, Florida residents over the age of 65. Th e White, non-Hispanic elderly population in Florida is much more evenly divided in terms of gender (male = 49%, female = 51%) than the study sample in which well over 50% of the sample is female (male = 37%, female = 63%). Further, study participants were also much more likely to live alone (47%) than other elderly Florid a residents (27%), less likely to be currently married and more commonly list themselves as previously married. Study respondents appear to be more highly educated, although comparisons ar e difficult to interpret since the data source consulted did not have educational st atus delineated by specific age groups. Nevertheless, with almost 61% of the study sample reporting edu cational achievement greater than 12 years, it is very likely th at the sample differs from the elderly Florida population in general. The study sample al so appears to demonstrate some health advantages over other elderly Floridians. Re spondents appear to be more likely to view
123 themselves as healthy or very healthy (69%) than their peers who are outside of the study, and report much lower levels of functional di fficulties than the general Florida population over 65. The two populations are closer t ogether, however, on the types and occurrence of chronic health conditions, with individuals in both groups reporting moderate to high levels of arthritis, hyperten sion, and heart disease.
124 Table 7 Comparing the Study Sample with White, Non-Hispanic Adults Aged 65 and Over in Florida _____________________________________________________________________________ Characteristic Percent ____________________________________________ Study Sample Florida Population (2000) ( N = 797) ( N = 2,807,597) ______________________________________________________________________________ Gender (male) 36.8 48.8 Gender (female) 63.1 51.2 Race (White) 100 78.0 Education >12 years 60.9 Marital Status Never Married 2.2 3.0 Married 52.3 60.6 Previously Married 45.4 36.6 Living Arrangements Living Alone 46.5 26.6 ______________________________________________________________________________ Continued on next page
125Table 7 (continued) ______________________________________________________________________________ Self Appraisal of Health Sick Â– Very Sick1 1.85 24.3 Fairly Healthy2 29.1 34.6 HealthyVery Healthy3 68.9 41.2 ADL Difficulties None 93.7 38.8* 1-2 ADLs 3.3 18.2 IADL Difficulty None 78.5 38.8* At least 1 IADL 8.53 13.6 Selected Chronic Conditions Arthritis 54.0 50.2 Hypertension 33.0 50.4 Heart Disease 24.9 33.4 ______________________________________________________________________________ Note: 1 corresponds to fair to poor. Note: 2 corresponds to good. Note: 3 corresponds to excellent to very good. Reported no ADL or IADL difficulties, but reported at least minimal physical limitations in other areas (e.g., stooping, lifting, grasping, walking distance).
126 Bivariate Analyses Previous research has often suggested that many of the variables commonly included in social and behavioral science studies are associated with gender. In an effort to frame findings in terms of some of the gender differences found in this study, initial analyses explored the association of gender w ith a number of the variables included in the analytical model. The Chi-square test of independence was used to estimate associations between the sociodemographic control variable s gender, living arrangement, and marital status. The Chi-square tested the null hypothe sis that in this st udy population gender was not significantly related to the categorical variables marital status and living arrangement. The larger the value of the Chi square, th e stronger the associ ation between the two variables in the sample being examined. The results of these analyses are presented in Table 5. As expected, gender was significantly associated with marital status and living arrangements. It has been well established in the United States that, in general, women tend to live longer than men (Hooyman & Kiya k, 1999). In light of that demographic trend, it follows that women are more likely to be widowed at higher rates, and live alone in greater numbers. The data for this sample appear to reflect this trend in that, women were half as likely to be married and more than twice as likely to live alone. In addition, to gain a more thorough description of the re ligious orientation of the study population, one-way ANOVAs were computed to explore the associations betw een gender and each of the variables that were combined to form the religiosity index. The results, which are summarized in Table 5 indicated that there we re significant differences in most of the religiosity domains measured in this study. The only exception was in the number of
127 religious activities that re spondents reported participating in. Female respondents scored significantly higher in each of the other religiosity domains. To test the associations between gender and continuous or inte rval level variables to be controlled at Time 1, one-way ANOVAs were computed. The one-way ANOVAs tested the hypothesis that in the sample, there were no differences in self-appraisal of health, positive or negative affect, base line life satisfaction levels or depressive symptomatology, IADL or ADL difficulties, or number of chronic conditions based upon gender. The results summarized in Table 5 indicate that statistically significant differences by gender were found for a number of measures. Women demonstrated higher mean levels of positive affect (16.31) than the men did (15.41) as well as higher levels of negative affect (8.98) than males (7 .66) Women also reported lower levels of baseline life satisfaction (18.45) than me n (19.22) in the study, and, in a finding somewhat consistent with previous studies women also reported significantly higher levels of baseline depressive symptomatology (7.62) than did the males (5.82). There were small, but statistically significant differences in levels of physical and functional health by gender as well, with women reported slightly higher levels of IADL difficulty and greater numbers of chronic conditions than did males. However, differences between males and females with regard to ADL difficulties and self-appraisal of health failed to reach significance. Current literature suggests that there may also be gender differences in the influence asserted by the independent variable (religiosity) included in this study. In an effort to more clearly describe the samp le population and to explore the relationship between gender and the individua l domains of religiosity that were combined to form the
128 independent variable in this study, one-wa y ANOVAs were computed. As summarized in Table 5, analyses revealed that gender wa s statistically associat ed with respondentsÂ’ reports of how religious they viewed themselves, with females reporting higher mean levels (Mean = 3.71) of self-assessed religiosity than male respondents. There were also significant differences in attendance of religious services, with females reporting higher levels of attendance. In regards to more s ubjective measures of religiosity or religious belief, females reported statistically higher leve ls of trust in God, ability to find comfort in religion and stronger tende ncies to seek GodÂ’s help when faced with stressful situations. Further, to determine the asso ciation between gender and religiosity, as measured by the religiosity index, a one-way ANOVA was computed. This analysis revealed that women demonstrated statistical ly higher mean scores (Mean = 3.66) on the index than did the men (Mean = 3.23). These results, which demonstrated that women score significantly higher than men on measures representing multiple domains of religiosity, are similar to those found by Idler, et al. (2003) in a natio nally representative sample. Spearman correlation coefficients were computed to assess the relationships between all control variables and baseline covari ates measured at interval or continuous levels. Results appear in Table 8. Results were generally as expected in terms of which variables demonstrated correla tions and the direction of th e correlations. However, not all correlations were significant, and many that were significant reflected only weak associations (e.g., .01 to .11). In general, the strongest associations were demonstrated among variables related to emotional and ps ychological well-being (e.g., positive affect, negative affect, life satisfaction, depre ssive symptomatology). Respondent age was
129 negatively associated with negative affect and positive affect at significant levels, and positively with IADLs, indicating that as st udy respondents increased in age, they were more likely to report higher levels of negative affect and IADL difficulties and lower levels of positive affect. Education was significantly associated with total assets and positive affect. Not surprisingly, total assets were positively associated at significant levels with life satisfaction, but negatively associated with negative affect, depressive symptomatology, IADLs, and number of chr onic conditions. These results are in line with numerous studies that have demons trated that individuals with greater socioeconomic resources tend to report higher levels of life satisfaction and positive affect, and are somewhat more likely to report greater levels of negative affect, depressive symptomatology, functional difficulties and health issues. As was anticipated, health and functi on variables were correlated with a large number of variables in the study. Health se lf-appraisal was nega tively correlated with negative affect, depressive symptomatology, IADLs, ADLs, and ch ronic conditions, but positively correlated with positive affect and life satisfaction. These relationships indicate that those respondents with lower levels of negative affect, depressive symptomatology, IADL and ADL difficulties, are more likely to view themselves as healthier. In addition to correlations with age and health self-image, IADLs were also correlated significantly with negative affect, life satisfaction, depressive symptomatology, ADLs and chronic conditions These results i ndicate that those respondents reporting greater leve ls of IADL difficulties were more likely to be older, report higher levels of negative affect, de pressive symptomatology, ADL difficulties, more chronic conditions and lower levels of life satisfaction. ADLs, although less
130 strongly associated with age than IADLs, were also significantly co rrelated with negative affect, depressive symptomatology, and chroni c conditions. Surprisi ngly, they were not significantly correlated with life satisfaction, which may reflect the fact that the study sample reported low levels of ADL difficulties at baseline. In regard to measures of emotional and psychological well-being, as anticipated, positive affect was significantly and positively correlated with education, health selfimage, life satisfaction, and re ligiosity, but negatively correl ated with age and depressive symptomatology. Whereas negative affect wa s significantly and negatively related to age, assets, health self-image, positive affect, and life satisfaction, but positively correlated with depressive symptomatology, IADLs, ADLs, and number of chronic conditions. Thus, respondents re porting greater levels of nega tive affect were more likely to be older, report fewer total assets, lower health self-image, positive affect and life satisfaction, but higher levels of depressive symptomatology, functional difficulties and chronic conditions. Depressive sympto matology was significantly and negatively correlated with assets, health self-image, positive affect, life satisfaction, and religiosity, but positively correlated with negative a ffect, IADLs, ADLs, and number of chronic conditions. Life satisfaction wa s also significantly correlated with a large number of the controls and other covariates in the study. Total assets, health se lf-appraisal, positive affect, and religiosity were positively correla ted with life satisfaction, whereas, negative affect, depressive symptomatology, IADLs and chronic conditions were negatively correlated. These results indicate th at respondents reporti ng higher levels of life satisfaction were more likely to report greater total assets, higher levels of positive affect and religiosity (independent variable), view themselves as healthier, and report lower
131 levels of negative affect, depressive sympto matology, functional difficulties, and chronic conditions. Relationships between the Control Variables, Covariates, Optimism, Time 10 Life Satisfaction and Time 10 Depressive Symptomatology Based upon the conceptual model de veloped to guide this study, sociodemographic characteristics and covariates measured at baseline were expected to be significantly correlated with the independe nt variable (religiosity), with Time 5 optimism (mediator), and with Time 10 life satisfaction and Time 10 depressive symptomatology (dependent variables). In addition, it was hypothesized that Time 5 optimism (mediator) would be significantly asso ciated with the religiosity (independent variable), and with Time 10 life satisfaction and Time 10 depressive symptomatology (dependent variables). Correlational analyses were computed to assess the strength of association of interval or continuous level control variables and covariates with the hypothesized independent variable (religios ity), mediator (optimism) and dependent variables (Time 10 life satisfa ction, Time 10 depressive symptomatology), as well as between the religiosity (independent variab le) and optimism (mediator), and between optimism and the dependent variables. Results are presented in Table 9. Correlational analyses revealed that, of the sociodemogra phic control variables, total assets had a significant relationship with both Time 10 life satisfaction and Time 10 depressive symptomatology (dependent variables), but not with optimism (mediator). Respondent age and level of education were not significan tly associated with the mediating variable
132 Table 8 Correlations Between Sociode mographic Variables, Covariates, and th e Religiosity at Baseline (N = 797) ________________________________________________________________________ ____________________________________ Variable 1 2 3 4 5 6 7 8 9 10 11 12 _____________________________________________________________________________________________________________________ 1. Age 1.00 -.02 .04 .02 -.11* -.14* .04 -.01 -.04 .07* .05 -.01 2. Education -.02 1.00 .15* .01 -.02 .09* -.02 -.03 -.04 .04 .05 -.01 3. Assets .04 .15* 1.00 .04 -.08* .02 .11* -.12* -.03 -.08* -.05 -.12* 4. Health Self-Appraisal .02 .01 .04 1.00 -.18* .17* .21* -.29* .06 -.28* -.13* -.41* 5. NA1 -.11* -.02 -.08* -.18* 1.00 -.01 -.33* .52* .02 .16* .09* .20* 6. PA1 -.14* .09* .02 .17* -.01 1.00 .19* -.30* .20* -.04 -.04 -.05 7. SWLS1 .04 -.02 .11* .21* -.33* .19* 1.00 -.45* .10* -.09* -.05 -.20* 8. Depressive Symptomatology -.01 -.02 -.12* -.29* .52* -.30* -.45* 1.00 -.10* .20* .08* .22* 9. Religiosity -.04 -.04 -.03 .06 .02 .20* .11* -.10* 1.00 -.04 -.04 -.04 10. IADLs .08* .04 -.08* -.28* .16* -.04 -.09* .20* -.04 1.00 .54* .32* 11. ADLs .05 .05 -.05 -.13* .09* -.04 -.05 .08* -.04 .54* 1.00 .22* 12. Chronic Conditions -.01 -.01 -.12* -.41* .20* -.05 -. 20* .22* -.04 .32* .22 1.00 ____________________________________________________________________________________________________________ Indicates statistical significance at p .05
133 or with either of the dependent variables. As expected, however, there were significant positive associations between optimism and the covariates health self-image (0.32), positive affect (0.23), baseline life satisfaction (0.26). There a significant positive association between the indepe ndent variable (religiosity) and optimism (0.17), although it was surprisingly weak. Also expected were the significant negative relationships demonstrated between Time 5 optimism and baseline measures of negative affect (-0.20), and depressive symptomatology (-0.36). T hus, respondents who reported a more positive health self-image, greater positive affect, and higher levels of life satisfaction at baseline also tended to report higher levels of optim ism at Time 5. Additionally, respondents who reported higher levels of religiosity (indepe ndent variable) at baseline also tended to report higher levels of optimism at Time 5. Surprisingly, optimism was not significantly associated with chronic conditions or IADL difficulties, both of which one might expect to challenge an individualÂ’s ab ility to remain optimistic or hopeful about the future. Optimism was, however, significantly asso ciated with both Time 10 depressive symptomatology (-0.36) and Time 10 life sa tisfaction (0.41), i ndicating that those respondents who reported greater depressive symptomatology at Time 10 were more likely to have reported lower levels of optimism at Time 5. Depressive symptomatology at Time 10 was positively correlated at significant levels with baseline measures of negativ e affect (0.41), depressive symptomatology (0.47), IADLs (0.15) and chronic conditions (0.16), indicating th at respondents who reported higher levels of depressive symp tomatology at Time 10 were more likely to have reported higher levels of negative affect, depressive symptomatology, IADL difficulties and chronic conditions at baseline Significant negative correlations were
134 found between baseline depressive symptomato logy, health self-image (-0.31), positive affect (-0.20), and life satisfaction (-0.30). Time 10 life satisfaction demonstrated significant relationships with the greatest number of baseline variables. Life satisfaction was positively associated with baseline measures of health self-image (0.23), positive affect (0.26), and life satisfaction (0.48), as well as religiosity (independent variable) (0 .23). Significant negative associations were found between Time 10 life satisfaction and base line negative affect (-0.27), depressive symptomatology (-0.37), IADLs (-0.13), ADLs (-0.14), and chronic conditions (-0.15). Thus, respondents who reported higher levels of life satisfaction at Time 10, tended to have reported higher levels of baseline heal th self-image and life satisfaction, fewer functional difficulties, lower depressive symptomatology, more positive mood, and greater religiosity at baseline. Correlations were also computed to test the relationship between the two dependent variables, Time 10 depressive sy mptomatology and Time 10 life satisfaction. The results indicated that these were the mo st strongly correlated study variables (-0.52), demonstrating that higher life satisfaction was associated with lower levels of depressive symptomatology.
135 Table 9 Correlations Between Control Variables, Co variates, Optimism, Time 10 Life Satisfac tion and Time 10 Depressive Symptomatology (N = 255) _______________________________________________________________________________________________________________________________ __ Variable Time 5 Optimism Time 10 Life Sati sfaction Time 10 Depressive Symptomatology (Baseline) (Mediator) (Dependent Variable) (Dependent Variable) _______________________________________________________________________________________________________________________________ __ Sociodemographic Controls Age -.002 .003 .11 Education -.11 -.06 -.07 Assets .06 .15* -.12* Covariates Health self-appraisal .32* .23* -.31* Negative affect -.20* -.27* .41* Positive affect .23* .26* -.20* Life satisfaction .26* .48* -.30* Depressive symptomatology -.36* -.37* .47* IADLs -.10 -.13* .15 ADLs -.12 -.14* .11 Chronic conditions -.11 -.15* .16 Independent Variable Religiosity .17* .23* -.07 Mediating Variable Time 5 Optimism 1.00 .41* -.36* Dependent Variables T10 Life satisfaction .41* 1.00 -.52* T10 Depressive symptoma tology -.36* -.52* 1.00 _______________________________________________________________________________________________________________________________ __ Note: (Sample N includes respondents who participated at Time 1, Time 5, and Time 10) = < .05
136 One-way ANOVAs were also computed to assess the associ ations between the categorical control variables gender, marital status and living arrangement, and optimism (mediator), Time 10 life satisfaction and Time 10 depressive symptomatology (dependent variables). Results are summarized in Table 10. No significant differences in levels of optimism were found based on gender, marita l status, or living arrangement. Not surprisingly, however, significant differences in mean levels of optimism, Time 10 life satisfaction, and Time 10 depressive sympto matology were associated with all three of these sociodemographic variables. Males we re found to have higher levels of life satisfaction and lower levels of depressive symptomatology than female respondents. Married respondents were more likely to re port higher levels of life satisfaction and lower levels of depressive symptomatology. In addition, respondents who indicated that they did not live alone at baseline reported significantly higher levels of life satisfaction and lower levels of depressive symptomatology at Time 10.
Table 10 Summary of ANOVA Results Assessing the Associations between Categorical Sociodemographic Cont rol Variables, Optimism, Time 10 Life Satisfaction and Time 10 Depr essive Symptomatology (N = 255) ____________________________________________________________________________________________________________ Optimism Life Satisfaction Depressive Symptomatology (Mediator) (Dependent Variable) (Dependent Variable) ____________________________________________________________________________________________________________ Mean (SD) F Mean (SD) F Mean (SD) F ____________________________________________________________________________________________________________ Gender Male 22.16 (2.33) 0.00 19.36 (3.60) 5.34* 8.34 (5.68) 9.57* Female 22.16 (2.44) 18.27 (3.59) 10.62 (5.60) ____________________________________________________________________________________________________________ Marital Status Never married 20.25 (2.96) 1.65 16.25 (4.80) 6.16* 10.89 (7.77) 4.79* Married 22.29 (2.32) 19.30 (3.51) 8.82 (5.31) Previously married 22.09 (2.43) 17.97 (3.52) 11.04 (5.90) ____________________________________________________________________________________________________________ Living Arrangement Alone 22.00 (2.48) 1.03 17.90 (3.64) 9.03* 11.10 (5.98) 10.89* Not alone 22.30 (2.32) 19.25 (3.52) 8.79 (5.93) ____________________________________________________________________________________________________________ Note: (Sample N of respondents who participat ed at Time 1, Time 5, and Time 10) Indicates statistical significance at p<.05
138 Multivariate Analysis Overview This study addressed two major hypotheses: (1) Religiosity is associated with subjective well-being in elderly adults, afte r controlling for sociodemographic factors, and baseline measures of physical health and function, positive and negative affect, depressive symptomatology, and life satisfactio n; (2) The association between religiosity and subjective well-being in elderly adults is mediated by optimism after controlling for sociodemographic factors, and baseline meas ures of physical health and function, positive and negative affect, depressive symp tomatology, and life satisfaction. For each hypothesis, subjective well-being at Time 10 was operationalized with the use of two measures: 1. Time 10 depressive symptomatology, as measured by a revised version of the Center for Epidemiologic Study of Depression (CES-D); 2. Time 10 life satisfaction, as measured by the Satisfaction with Life Scale (SWLS). The multivariate model building process was informed by research strategies outlined by Baron and Kenny (1986) and guided by the conceptual graphic included in Chapter 1. Utilizing the hierarchical regressi on strategy summarized in Figure 8, a series of steps were built upon one anot her to create the final regr ession model. Step 1 was a two-part process in which each of the dependent variables (Time 10 depressive symptomatology, Time 10 life satisfaction) wa s first regressed on the control variables that had been entered as a block, then regres sed on the block of control variables with the
139 covariates added to the m odel individually. To explore the association between the independent variable (Time 1 religiosity) a nd the dependent variables, Step 2 involved regressing the dependent variables on the inde pendent variable. At Step 3, each of the dependent variables was regressed on the mediator (Time 5 optimism). Step 4 assessed the predicted relationship betw een the mediator (optimism) and the independent variable (religiosity). Since it was predicted that re ligiosity at Time 1 would predict optimism scores at Time 5, optimism was regressed on religiosity. For Step 5, the dependent variables were regressed on a linear combinati on of the control variables, the covariates, and the independent variable (religiosity ). To assess the proposed mediational relationship between optimism and the depende nt variables, Step 6 entailed adding optimism to models in which the dependent variables were regressed on the controls, covariates, and independent (relig iosity) variable. Each of thes e steps, and the results they generated, are discussed in the following sections. Step Variables Included in Model 1. Regression of dependent variables on: a) Control variables as a block b) Control variables (block) + covariates (sequenced) 2. Regression of optimism (mediator) on: Religiosity (independent variable) 3. Regression of dependent variables on: Control variables (block) + covariates (block) + religiosity (i ndependent variable) 4. Regression of dependent variables on: Control variables (block) + covariates (block) + religiosity + optimism (mediator) Figure 9. Steps in Building the Hierarchical Regression Model Predicting Time 10 Depressive Symptomatology and Life Satisfaction
140 Regression Diagnostics The effects of outliers on the regression line were explored through inspection of residuals (RStudent), influence (CookÂ’s D), leve rage (hat values), diffits, and effects on specific parameter estimates (dfbetas). An ou tlier is an unusual or distinctly different data point that is distinct from other data points (Pedhazur & Schmelkin, 1997). Outliers may result from a number of situations includ ing, input errors, meas urement errors, poor instrumentations, to name a few. However, outliers may also occur in the absence of errors (Pedhazur, 1997) and may not necessa rily warrant deletion from the dataset. Examination of residual plots indicated relativ ely normal distribution, absent of nonlinear or heteroscedastic patterns. Criteria for extreme influence included, cases with a CookÂ’s D greater than 1 or cases that demonstrated two or more of the following: a) a hat value > twice the mean hat value for the sample; b) an RStudent >3; c) a DFBETA >2; or d) a DFFITS > 2( k/n). There were no cases that met these criteria. Collinearity diagnostics indi cated that the overlap of variables in the model was considerably less than the 80% identified by Pedhazur and Schmelkin (1997) as problematic. Tolerance estimates for the inde pendent variables were greater than 0.2. Variance inflation factor estimates all fell under 0.4. Thus, no variables were removed from the model due to multico llinearity effects. Building the Regression Model For Step 1, each of the dependent variables, Time 10 depressive symptomatology and Time 10 life satisfaction, were regre ssed on the linear combination of the seven control variables. A summary of the results is provided in Table 11 When entered as a
141 block, the control variables accounted for approximately 19 % of the observed variance in Time 10 depressive symptomatology (F[7, 247] = 8.07, p<.01, adjusted R2 = .16) The block of control variables accounted for approxi mately 15% of the observed variance in Time 10 life satisfaction (F[7, 247] = 6.48, p<.05, adjusted R2 = .13). Results for Step 1 indicate that the cont rol variables alone account for a mo derate proportion of the total variance observed in the depende nt variables. All associations were in the expected direction and were consistent with bivariate statistical results. Also consistent with previous literature, health self-image ( b = -2.67, t = -5.98, p<.05), gender ( b = 1.57, t = 2.10, p<.05) and living arrangements ( b = 2.21, t = 2.20, p<.05) appear to contribute the most to the explanation of the variance in Time 10 depressive symptomatology. The parameter estimates ( b ) for all three of the variables were significant at the p<.05 level. Health self-appraisal also c ontributed significantly (b = 1.50, t = 5.20, p<.001) to the explanation of Time 10 life sati sfaction, as did living status Living alone was negatively associated with life satisfaction ( b = -1.65, t = -2.55, p <.05), whereas higher levels of health self-appraisal were predictive of hi gher levels of Time 10 life satisfaction. Somewhat surprising in these results is the fa ct that neither educa tion nor total assets were significant in the model. Much resear ch has suggested that both education and financial assets contribute to more positive physical and psychological outcomes.
142 Table 11 Regression Coefficients Obtained in Mult iple Regression Analyses with Control Variables Predicting Time 10 Depressive Symp tomatology and Life Satisfaction Â– Model 1 Control Variables (N = 255) Time 10 Depressive Symptomatology Time 10 Life Satisfaction Predictor b coefficient t b coefficient t Age 0.15 0.09 1.50 -0.01 -0.01 -0.10 Gender 1.57 0.13 2.10* -0.58 -0.08 -1.21 Marital Status -0.78 -0.07 -0.85 0.85 0.13 1.43 Living Status 2.21 0.19 2.20* -1.65 -0.23 -2.55* Health Appraisal -2.67 -0.34 -5.98* 1.50 0.31 5.20* Education -0.17 -0.74 -1.28 -0.05 -0.04 -0.63 Assets -0.06 -0.03 -0.59 0.12 0.11 1.85 Model Fit Statistics Time 10 Depressive Symptomato logy Time 10 Life Satisfaction Model R2 .019 0.15 Model F 8.07* 6.48* Indicates statistical significance at p<.05
143 Completing Step 1 Â– The Addition of Each Covariate to the Model To complete Step 1, the covariates were added individually to the control model. Parameter estimates ( b ), standardized multiple regression coefficients ( ), and the uniqueness index for each variable as it is added to the model are provided in Tables 12 and 13. Uniqueness indices were tested for statis tical significance with an F test which tests the null hypothesis that th e uniqueness index for the variab le of interest is equal to zero. The order in which the covariates were adde d to the model was based upon the magnitude of influence each was expected to contribute to the explanatory ability of the model. Those variables that were expected to be most influe ntial were added first. For example, positive and negative affect have been shown to be very clo sely related to the dependent variables, and in some cases have been considered as proxy measures for them. Likewise, baseline measures of life satisfaction and depressive symptomatology were expected to be moderately predictive of the dependent variables. Indeed, all four variables had been found to be moderately and significantly correlated with the dependent variables in biva riate analyses. The remaining health and function variables were expected to exert relatively less explanatory power to the model, and thus were added later in the model.
144 Table 12 Building Model 1 Â– Regression Coefficients and Uniqueness Indices Obtained in Multiple Regression Analyses Predicting Depressiv e Symptomatology Controls + Individual Covariates (N = 255) Predictor b coefficient t Model R2 Uniqueness Index Model F Control Variables .19 8.07* Controls + Covariates .33 8.60* Negative Affect 0.52 0.36 4.98* .26 .07* 10.84* Positive Affect -0.33 -0.21 -3.83* .30 .04* 11.81* Life Satisfaction -0.15 -0.09 -1.41 .31 .01* 10.87* Depressive Symptomatology 0.24 0.21 2.80* .33 .02* 10.87* IADLs -1.04 -0.04 -0.81 .33 .002 10.00* ADLs 4.43 0.09 0.02 .33 .002 9.29* Chronic Conditions -0.07 -0.02 -0.34 .33 .00 8.60* Indicates statistical significance at p<.05
145 Table 13 Building Model 1Regression Coefficients and Uniqueness Indices Obtained in Multiple Regression Analyses Predicting Life Satis faction Â– Controls + Individual Covariates Predictor b coefficient t Model R2 Uniqueness Index Model F Control Variables .15 6.48* Controls + Covariates .34 9.03* Negative Affect -0.26 -0.23 -3.77* .20 .05* 7.75* Positive Affect 0.22 0.22 3.92* .25 .04* 9.00* Life Satisfaction 0.38 0.35 5.70* .34 .09* 12.38* Depressive Symptomatology -0.07 -0.10 -1.34 .34 .004* 11.46* IADLs -0.15 -0.01 -0.19 .34 .00 10.46* ADLs -3.38 -0.11 -1.11 .35 .00 9.76* Chronic Conditions 0.03 0.01 0.22 .35 .00 9.03* Indicates statistical significance at p<.05
146With only the control variables in Model 1 (Table 11), gender, living status and health self appraisal were significan tly associated with Time 10 depressive symptomatology. With the addition of negative affect to th e model, gender became insignificant, but age became significant, and remained so throughout the addition of the remaining covariates. The lack of significance of gender in this model was somewhat unexpected given the significant findings in both prev ious literature and the bivari ate analyses for this study, which indicate that levels of depressive symptomatology tend to vary by gender. The next covariates added to the model, positive a ffect, followed by baseline life satisfaction, revealed significant associations with th e dependent variable. Although demonstrating significance itself, once baseline depressive symptomatology was added to the model, living status was no longer significant. Howe ver, living status became significant again when IADL measures were added to the model, and remained significant with the addition of the remaining covariates (ADLs, chronic conditions). Again, somewhat surprising were the lack of significant effects indicated when the health and physical effects variables were added to the model. Despite significant correlational relationships with Time 10 depressive symptomatology, neither IADLs (r = -.15, p<.05) nor chronic conditions (r = .16, p<.05) reached signi ficance at any time in the model. As expected, once all the covariates were a dded to the control model, a much greater proportion of the variance was explained for each of the dependent variables. As a linear combination, the controls and covariates account ed for approximately 33% of the variance in Time 10 depressive symptomatology, whereas the combination accounted for approximately 34% of the variance in Time 10 life satisfaction. Unexpectedly, however, few of the associations between the covariates and the dependent variabl es reached significance. In predicting Time 10
147depressive symptomatology, only positive affect ( b = -.23, t = -2.51, p<.05) and baseline depressive symptomatology ( b = .25, t = 2.87, p<.05) reached significance. Based upon bivariate results, it had been anticipated th at baseline life satisfaction (r = -.30, p<.01)) and negative affect (r = .41, p<.01) would show significant predictive influence as well. Similarly, only positive affect ( b = .13, t = 2.24, p<.05) and baseline life satisfaction ( b = .37, t = 5.49, p<.01) exhibited significant influence in the explanation of variance in Time 10 life satisfaction. Based upon bivariate results, it was expected that baseline depressive symptomatology (r = .37, p<.01) and negative affect (r = -.27, p<.01) would be signifi cant predictors of Time 10 life satisfaction. Table 14 shows the results of Step 1 when the model was run with controls and covariates added simultaneously to predict Time 10 depressive symptomatology and Table 15 contains the results for the prediction of Time 10 life satisfaction wh en the controls and covariates were entered simultaneously. When Time 10 life satisfaction was regresse d on the control variables entered as a block, approximately 15% of the variance was explained. In this initial model, living status ( b = -1.65, t = -2.55, p<.05) and health self appraisal ( b = 1.50, t = 5.20, p<.01) were significant. Although they remained significant throughout the addition of the covariates, their influence was reduced furthe r as each covariate was added to the model. Negative affect ( b = -.26, t = -3.77, p<.05), positive affect ( b = .22, t = 3.93, p<.05), and baseline life satisfaction ( b = .38, t = 5.70, p<.01) all demonstrated significance as they were added to the model. However, nega tive affect was no l onger significant once baseline depressive symptomatology was added to the model. As suggested in some previous studies, this could be a reflection of overlapping influence of negative affect and depressive symptomatology. Not surprisingl y, as the physical health and functioning variables were added to the model, the infl uence of baseline depressive symptomatology,
148 though still not significant, an d life satisfaction increased. Thus, in predicting Time 10 life satisfaction among this sample, respondent physical health and functioning seem to contribute to the predictive power of bot h baseline depressive symptomatology and baseline life satisfaction. The final task of Step 1 was to regr ess each of the dependent variables on the controls and covariates that had been adde d into the model simultaneously. These results are summarized in Table 16. Results indicate that the linear combination of all the controls and covariates accounts for appr oximately 33% of the variance in Time 10 depressive symptomatology and 34% of the va riance in Time 10 life satisfaction. Once the controls and covariates were entered together as a block, results were more consistent with expectations. For example, in predicting Time 10 depressive symptomatology, living status ( b = 1.92, t = 2.02, p<.05) and health self-appraisal (b = -1.86, t = -.0.24, p<.05) were significant, indicat ing that living alone is pred ictive of higher levels of depressive symptomatology, and that those wi th a more negative health self-appraisal are more likely to demonstrate higher levels of depressive symptomatology. Also significant in this model was respondent age ( b = 0.21, t = 2.31, p<.05) indicating that, as age increases, depressive symptomatology is mo re likely to increase as well. These relationships are all consistent with previous research with elderly populations. Also, as might be expected, other predictors demons trating significant influence on depressive symptomatology at Time 10, were base line measures of positive affect ( b = -0.23, t = 2.51, p<0.5) and depressive symptomatology ( b = .25, t = 2.87, p<.05), both of which had demonstrated significant correlation with Time 10 depressive symptomatology in bivariate analyses. These results suggest that higher levels of positive affect are
149 associated with lower levels of depressive symtomatology, and that respondents reporting higher levels of depressive symptomatology at baseline, are more likely to report higher levels of depressive symptomatology at Time 10. Baseline measures of negative affect and life satisfaction, however, both of whic h had demonstrated significant bivariate correlations with Time 10 depressive symp tomatology, were not significant in these analyses.
150 Table 14 Model 1 Regression Coefficients Obtained in Multiple Regression Analyses Predicting Depressive Symptomatology with Controls and Covariates (N = 255) Time 10 Depressive Symptomatology Time 10 Life Satisfaction Predictor b coefficient t b coefficient t Age 0.21 0.12 2.31* -0.04 -0.04 -0.75 Gender 1.21 0.10 1.66 -0.53 -0.07 -1.15 Marital Status -0.99 -0.09 -1.15 0.74 0.11 1.38 Living Status 1.92 0.17 2.02* -1.25 -0.17 -2.09* Health Appraisal -0.12 -0.05 -1.00 0.93 0.19 3.06* Education -1.86 -0.24 -3.86* -0.0 -0.05 -1.01 Assets -0.04 -0.02 -0.45 0.06 0.06 1.06 Negative Affect 0.24 0.13 1.83 -0.01 -0.01 -0.18 Positive Affect -0.23 -0.15 -2.51* 0.13 0.13 2.24* Life Satisfaction -0.11 -0.07 -1.07 0.37 0.34 5.49* Depressive Symptomatology 0.25 0.25 2.87* -0.08 -0.10 -1.42 IADLs -2.78 -0.12 -1.17 1.20 0.08 0.80 ADLs 4.48 0.09 0.92 -3.41 -0.11 -1.11 Chronic Conditions -0.07 -0.02 -0.34 0.03 0.01 0.22 Model Fit Statistics Model R2 .33 .34 Model F 8.60* 9.03* Indicates statistical significance at p<.05
151 Step2 Â– Adding Religiosity to the Model For Step 2, the dependent variables were regressed on a linear combination of the controls and covariates, which were added as a block, and the independent variable (religiosity), which was added last to the model. Contrary to expect ations, the addition of religiosity to the model did not increase th e proportion of variance explained in Time 10 depressive symptomatology. As seen in Table 17, the addition of religiosity to the second model had very little impact on the pa rameter estimates or beta weights for any of the variables that had been entered into the fi rst model. Further, the uniqueness index for religiosity was insignifican t (U = .00002380). Although relig iosityÂ’s association with Time 10 depressive symptomatology was in the expected direction (n egative), religiosity actually accounted for the least amount of influe nce of any of the predictors in the model. Thus, although Model 2 explained a signifi cant proportion of the variance in Time 10 depressive symptomatology, the addition of re ligiosity to the model did not result in a better explanatory model than the first model. Therefore, hypothesis 1.2, that religious involvement is inversely associated with Time 10 depressive symptomatology, after controlling for sociodemographic factors and co variates, was not sta tistically supported. Further, in light of the fact that religios ity was not a significant predictor of Time 10 depressive symptomatology (dependent vari able), conditions necessary to test for mediation were not met. Therefore, Hypothe sis 2.2, which posited that optimism would mediate a significant relationship between religiosity and Time 10 depressive symptomatology once sociodemographic variable s and covariates were controlled, was not supported. The results of Model 2 in re gard to Time 10 depressive symptomatology
152 suggest that: a) as age increases, depressive symptomatology is more likely to increase; b) living alone at baseline is more likely to be associated with higher levels of depressive symptomatology at outcome; c) health self-app raisal is negatively related to depressive symptomatology, in that those who view them selves as average to moderately healthy report lower levels of depressive symptomato logy; d) individuals reporting higher levels of depressive symptomatology at Time 1 were more likely to report higher levels of depressive symptomatology at Time 10. Su rprisingly, gender did not appear to be significantly associated with Time 10 depres sive symptomatology, nor did marital status, assets or life satisfaction, despite significan t bivariate correlations and assertions by many researchers of their significant influence on depr ession. Also, neither levels of functional disability nor number of chr onic conditions proved significan t despite expectations that, in older populations, functional decline norma lly associated with age might lead to greater levels of depressive symptomatology. However, as previously stated, this sample was drawn from an elderly population which se ems to enjoy better than average health, education, and financial security. When religiosity was added to the model predicting Time 10 life satisfaction, however, there was a significant association (Table 15). When religiosity was added to the control variables and cova riates in Model 2, there was a significant increase in the proportion of variance explained and in the Model F value ( R2 = .37, F[15, 239], p<.001). Health self-appraisal and baseline life satisfaction demonstr ated significance, as they had in the first model, but positiv e affect and living status were no longer significant. Thus, in the prediction of Time 10 life satisfaction, Model 2 demonstrates that a) positive health self-appraisal predicts higher levels of life satisfaction; b) those
153 respondents with higher levels of life satisfaction at Time 1 were more likely to report higher levels of life satisfaction at Time 10; c) religiosity was positively associated with Time 10 life satisfaction, in that those respondents who reported higher levels of religiosity at baseline, were significantly more likely to report higher levels of life satisfaction at Time 10. Therefore, Model 2 demonstrated support of hypothesis 1.1, that religious involvement is positively associated with Time 10 life satisfaction after controlling for sociodemographic factors and covariates.
154 Table 15 Model 2 Predicting Time 10 Depressive Sy mptomatology and Life Satisfaction Â– Regression Coefficients Obtained in Multip le Regression Analyses with Controls, Covariates, and Religiosity (N = 255) Time 10 Depressive Symptomatology Time 10 Life Satisfaction Predictor b coefficient t b coefficient t Control Variables Age 0.21 0.13 2.31* -0.05 -0.05 -0.71 Gender 1.22 0.10 1.65 -0.76 -0.10 -1.67 Marital Status -0.98 -0.09 -1.13 0.54 0.08 -1.88 Living Status 1.92 0.17 2.00* -1.11 -0.15 1.01 Education -0.12 -0.05 -0.99 -0.08 -0.05 -1.04 Health Appraisal 11.86 -0.24 -3.86* 0.95 0.19 3.20* Assets -0.04 -0.02 -0.45 0.07 0.06 1.14 Covariates Negative Affect 0.24 0.13 1.82 -0.03 -0.03 -0.43 Positive Affect -0.23 -0.14 -2.44* 0.09 0.09 1.58 Life Satisfaction -0.11 -0.07 -1.05 0.35 0.33 5.26* Depressive Symptomatology 0.25 0.21 2.84* -0.06 -0.08 -1.11 IADLs -2.76 -0.12 -1.16 0.95 0.06 -0.06 ADLs 4.47 0.09 0.92 -3.17 -0.10 -0.65 Chronic Conditions -0.07 -0.02 -0.09 0.04 0.02 -1.06 Independent Variable Religiosity -0.04 -.01 0.09 0.81 0.18 3.15* Continued on next page
155 Table 15 Continued Model Fit Statistics Time 10 Depressive Symptomatology Time 10 Life Satisfaction Model R2 .33 .37 Model F 8.00* 9.41* Indicates statistical significance at p<.05
156 Step 3 Â– Testing for Mediational Relationships Data analyses in Steps 1 and 2 addr essed the first summary hypothesis that religiosity is associated with subjective well-being when the linear combination of control variables and covariates are held constant. The analyses demonstrated that, for this sample, religiosity was significantly associated with Time 10 life satisfaction, but was not significantly associated with Ti me 10 depressive symptomatology. These analyses address the first of four conditi ons that, according to Baron and Kenny (1986), must be met in order to test for a mediat ion effect. To meet the first condition, a significant relationship must exist between the independent variable (religiosity) and the dependent variables (Time 10 life satisfacti on; Time 10 depressive symptomatology). Table 16 presents a summary of bivariate correl ations among the independent variable (religiosity), the proposed mediator (optimis m) and both dependent variables (Time 10 depressive symptomatology, Time 10 life sati sfaction). These result s indicate a weak and nonsignificant negative correlation betw een religiosity and Time 10 depressive symptomatology, and a relatively larger, significant correlati on with Time 10 life satisfaction..
157 Table 16 Correlational Relationships Among Independent, Depe ndent and Proposed Mediati ng Variables (N = 255) ____________________________________________________________________________________________________________ Variable Time 1 Time 10 Time 10 Time 5 Religiosity Depressive Life Satisfaction Optimism Symptomatology ____________________________________________________________________________________________________________ Religiosity 1.00 -.07 .23* .17* Depressive Symptomatology -.07 1.00 -.52* -.36* Life Satisfaction .23* -.52* 1.00 .41* Optimism .17* -.36* .41* 1.00 ____________________________________________________________________________________________________________ Indicates statistical significance at p<.05
158 Steps 3 and 4 examined the second and th ird conditions of mediation as suggested by Baron and Kenny (1986). Step 3 determ ined there was a significant relationship between the independent variable (religio sity) and the proposed mediating variable (optimism) by regressing optimism on religios ity, but once the controls and covariates were added to the model, the relationship was no longer significant. Step 4 determined there was a significant relationship between th e mediator (optimism) and the dependent variable (Time 10 life satisfaction) by re gressing life satisfaction on optimism. Step 5 tested the full model predicting Time 10 life satisfaction, adding optimism (mediator) to Model 3, in which Time 10 life satisfacti on (dependent variable) was regressed on a linear combination of the control variables, covariates and religiosity (independent variable). Table 10 summarizes the result s of regression analyses representing the conditions outlined in Baron and KennyÂ’s (1986) causal steps method of mediation analyses. The summary of regression analyses results indicates that two of the conditions for mediation were not met: (1) The independent variable (religiosity) was not significantly associated with Time 10 depressive symptomatology (dependent variable); (2) The independent variable (religiosity) was not significantly associated with the proposed mediator in the multivariate model. Thus, hypotheses 2.1 and 2.2, that optimism would mediate the association between religiosity and Time 10 depressive sympto matology and Time 10 life satisfaction were not supported. However, the results of the fina l model presented in Ta ble 17, indicate that the addition of optimism (mediator) to the li near combination of variables predicting Time 10 life satisfaction did result in a slight decrease in the influence of religiosity.
159 Therefore, a Sobel Test was calculated to test whether the indirect effect of the independent variable (religiosity) on Time 10 life satisfaction (dep endent variable) via the mediator (optimism) was significantly di fferent from zero. The calculation produced a Sobel test statistic of 1.49 (p value = 0.14), indicating no support for significant mediation effects.
160 Table 17 Model 3 Testing for Mediati onal Relationships (N = 255) ____________________________________________________________________________________________________________ Condition Tested etas (t Value) IV MV MV DV IV DV IV DV MVd With Controls and Covariates ____________________________________________________________________________________________________________ 1. Religiosity (IV)a Optimism (MV)b 0.18 ( 2.97*) 0.10 (1.58) 2. A. Optimism (MV) Depressive Sy mptomatology (DV)c -0.35 (-5.91*) -0.13 (-2.12*) B. Optimism (MV) Life Satisfaction (DV) 0.45 (8.12*) 0.27 (4.60*) 3. A. Religiosity (IV) Depressive Symptomatology (DV) -0.06 (-.96) -.005 (-.09) B. Religiosity (IV) Life Satisfaction (DV) 0.24 (3.89*) 0.18 (3.15*) 4. A. Religiosity (IV) Depressive Sy mptomatology (DV) (MV)d 0.01 (.12) B. Religiosity (IV) Life Satisfaction (DV) (MV)d 0.15 (2.81*) ____________________________________________________________________________________________________________ a IV = Independent Variable. b MV = Mediator Variable. c DV = Dependent Variable. d Controlling for sociodemographi c variables and covariates. Indicates statistical significance at p<.05.
161 Summary of Findings The first hypothesis for this study posite d that baseline measures of religiosity would be positively associated with Time 10 life satisfaction (hypothesis 1.1) and inversely associated with Time 10 depressive symptomatology (hypothesis 1.2) in elderly adults. The analyses demonstrated a signifi cant positive relationship between religiosity ( = 0.15, t = 2.81, p<.05) and Time 10 life sa tisfaction, but the relationship with Time 10 depressive symptomatology ( = 0.01, t = .10), although in the expected direction, was not significant. The beta weights for all cont rols and covariates were in the expected direction, though surprisingly, many did not de monstrate significance. Of the control variables and covariates included in th e models, only health self-appraisal ( = 0.14, t = 2.32, P<.05) and baseline life satisfaction ( = 0.30, t = 5.03, p<.05) demonstrated significant influence in predicting life satisfactio n, with levels of life satisfaction reported at baseline demonstrating the stronges t predictive power. Respondent age ( = 0.12, t = 2.24, p<.05), health self-appraisal ( = -0.21, t = -3.34, p<.05), positive affect ( = -0.14, t = -2.33, p<.05) and baseline depressive symptomatology ( = 0.18, t = 2.37, p<.05) were significant predictors of Time 10 depressi ve symptomatology. Thus, the strongest predictor of Time 10 depressive symptomato logy in these elderly respondents was the level of depressive symptomatol ogy they reported at baseline. The second set of hypotheses which were related to the mediation effects of optimism, were not supported. Consistent with recommendations of Baron and Kenny (1986), causal steps mediational analysis was not completed since the independent variable (religiosity) was not significan tly associated with Time 10 depressive symptomatology or the hypothesized mediator (optimism). Thus, hypotheses 2.1 and 2.2
162 were not supported. Although the addition of optimism to the final model suggested possible indirect effects in that the influe nce of religiosity on Time 10 life satisfaction was reduced slightly, a Sobel Te st indicated that the indirect effects of optimism were not significant. A summary of the contribution to th e final regression model made by each predictor is presented in Table 21. Table 22 presents a summary of the beta weights and t values for the significant predictors of each of the dependent variables in order of the strength of their influence. Figure 10 presents a summary of the findings demonstrated by the data analyses in relation to each of the hypotheses proposed by this study.
163 Table 18 Summary of Regression Analysis for Va riables Predicting Time 10 Depressive Symptomatology and Time 10 Life Satisfaction (N = 255) __________________________________________________ ______________________ Depressive Symptomatology Life Satisfaction R2 = .35 F = 7.89* R2 = .42 F = 10.67* ________________________________________________________________________ Predictor t t Age 0.12 2.24* -0.04 -0.74 Gender 0.11 1.74 -0.11 -1.90 Marital Status -0.07 -0.85 0.03 0.45 Living Status 0.15 1.78 -0.11 -1.48 Education -0.06 -1.15 -0.04 -0.77 Health Self Appraisal -0.21 -3.34* 0.14 2.32* Assets -0.02 -0.31 0.05 0.89 Negative Affect 0.12 1.71 -0.01 -0.19 Positive Affect -0.14 -2.33* 0.08 1.39 Life Satisfaction -0.05 -0.86 0.30 5.03* Depressive Symptomatology 0.18 2.37* -0.02 -0.26 IADLs -0.10 -0.98 0.03 0.29 ADLs 0.08 0.81 -0.08 -0.87 Chronic Conditions -0.02 -0.29 0.01 0.17 Religiosity 0.01 0.12 0.15 2.81* Optimism -0.13 -2.11* 0.25 4.36* ________________________________________________________________________ Indicates statistical significance at p<.05
164 Table 19 Summary Table of Beta Weights of the Mo st Influential Predictors of Time 10 Life Satisfaction and Time 10 Depressi ve Symptomatology (N = 255) Most Influential Predictors of Time 10 Life Satisfaction t Baseline Life Satisfaction 0.30 5.03* Optimism 0.25 4.36* Religiosity 0.15 2.81* Health Self-Appraisal 0.14 2.32* Most Influential Predictors of Time 10 Depressive Symptomatology t Health Self-Appraisal -0.21 -3.34* Baseline Depressive Symptomatology0.18 2.37* Positive Affect -0.14 -2.33* Optimism -0.13 -2.11 Age 0.12 2.24* Indicates statistical significance at p<.05
165 Study Hypotheses Findings 1. Religious involvement is positively associated with subjective well-being in elderly adults, after controlling for sociodemographic factors, baseline health and function, depressive symptomatology, life satisfaction, and positive and negative affect. This summary hypothesis was not completely supported. The findings were mixed. Religiosity, as measured at Time 1, was significantly associated with life satisfaction at Time 10, but was not significantly associated with depressive symptomatology at Time 10. 1.1. Religious involvement is positively associated with life satisfaction in elderly adults after controlling for sociodemographic factors, baseline health and function, depressive symptomatology, life satisfaction, and positive and negative affect. This hypothesis was supported. Time 1 religiosity was positively associated with Time 10 life satisfaction in a sample of elderly adults. Beta weights indicate that religiosity ( = .18, t = 3.15, p<.05) ranked third in strength of influence among the three variables that demonstrated significance in the regression model to predict life satisfaction. Baseline life satisfaction ( = .33, t = 5.26, p<.05) and health self-image ( = .19, t = 3.20, t = <.05) each demonstrated larger beta weights. 1.2. Religious involvement is inversely associated with depressive symptomatology in elderly adults after controlling for sociodemographic factors, baseline health and function, depressive symptomatology, life satisfaction, and positive and negative affect. This hypothesis was not supported. Although its beta weight was in the expected direction, religiosityÂ’s influence was weak and insignificant ( = -.01) in the prediction of depressive symptomatology in elderly adults. 2. The association between religious involvement and subjective well-being in elderly adults is mediated by optimism, after controlling for sociodemographic factors, baseline he alth and function, depressive symptomatology, life satisfaction, and positive and negative affect. This summary hypothesis was not supported. Conditions for mediation were not met. Figure 10. Summary of Findings By Study Hypothesis
166 Figure 10 ( Continued ). Study Hypotheses Findings 2.1. The association between religious involvement and life satisfaction in elderly adults is mediated by optimism, after controlling for sociodemographic factors, baseline health and function, depressive symptomatology, life satisfaction, and positive and negative affect. This hypothesis was not supported. When optimism (mediator) was added to the model, the effect of religiosity (independent variable) was slightly reduced, but remained significant in the explanation of varian ce in the dependent variable (Time 1 life satisfaction). However, a Sobel Test indicated that the indirect effect of religiosity (independent variable) Time 10 life satisfaction via optimism (mediator) was not significantly different from zero. 2.2. The association between religious involvement and depressive symptomatology in elderly adults is mediated by optimism, after controlling for sociodemographic factors, baseline health and function, depressive symptomatology, life satisfaction, and positive and negative affect. This hypothesis was not supported. Religiosity did not demonstrate a significant relationship with Time 10 depressive symptomatology, thus the criteria for mediation was not met and mediational analyses were not completed.
167 Chapter Five Discussion It has been well established in the litera ture that adults over the age of 65 are among the fastest growing segments of the popul ation in the United States. Consistent with this trend, researchers in the field of gerontology appear to be broadening their focus to look beyond the goal of increas ing the lifespan and are searching in earnest for ways to enhance the quality of later years. Resear chers have thus far failed to agree upon what, exactly, constitutes successful negotiation of the later years of the lifespan. Labels such as resilient, robust, productive, and hardy have been used to describe individuals who are able to successfully adapt to or rise above the challenges often created by the aging process. To date, there is no universally accepted definition or model of su ccessful aging nor is there consensus regarding how the con cept should be operati onalized. However, despite the fact that no common nomenclatu re or all inclusive theory have been established, a common theme throughout the fiel d of gerontology is th e need to identify personality, physical, social, and contextual fact ors and traits that di stinguish those older adults who age more successfully than those who do not. There has been a growing interest over the past decade in research aimed at linking religion or religious involvement with a broad ra nge of physical and mental health outcomes. There have been hundreds of empirical studies that have investigated the effects of some aspect of religion on h ealth, health behaviors, morbidity, mortality, and numerous psychological outcomes (e. g., Ellison, 1994; Koenig, 1990; Levin &
168 Schiller, 1987; Witter, Stock, Okun, & Haring, 1985). Despite limitations imposed by methodological differences and lack of a co mmon measurement strategy, the findings have consistently, though not universally, f ound evidence for the beneficial effects of religion (Levin, 1994a). Levin (1994a) has s uggested that religion serves as a profoundly important personal resource for older adults in particular, and that as individuals age, religiosity becomes strongly predictive of be tter health and well-being. Not only does the predictive power remain even after controlli ng for the effects of h ealth status, but the magnitude of the effect has been shown to be at least as great as that of health itself on well-being (Levin, 1994a). Levin has argued that this is particularly noteworthy in light of the fact that it has been generally accepted that oneÂ’s health is the best predictor of oneÂ’s well-being. Further, the influence of religion on well-being is not eliminated by controlling for sociodemographic factors. The purpose of this study was to determ ine if, while holding a number of possible confounders and covariates constant, religiosity predicts subjective well-being in elderly adults and, if it does, to explore the extent to which that relationship is mediated by dispositional optimism. The exploration of these possible resiliency resources is important for increasing the unde rstanding of how they might se rve to enhance successful adaptation to the changes inherent in ag ing and for gaining a better understanding of individual differences in la te life outcomes. Although th ese resources cannot stop the aging process, nor can they prevent loss or decline, they may be important tools for helping elderly adults to maintain their level of functioning, to recover more quickly or even, perhaps, advance beyond their previous level of functioning when confronted with challenges or adversity. The ability to pr otect against negative declines and losses
169 associated with aging and the promotion of resiliency and optimal adjustment in advancing years have the potenti al to increase the numbers of elders who are able to remain independent in the community as they successfully negotiate the aging process. Summary of Findings Hierarchical regression analyses were used to test hypotheses postulating that, with confounders and covariates controlled, re ligiosity is significan tly associated with depressive symptomatology and life satisfacti on in a sample of elderly, retirementcommunity-dwelling adults, and that those rela tionships are mediated by optimism. The hypotheses were not entirely supported by the an alyses. Results revealed that there was no evidence supporting the hypothesis that ther e is a significant association between religiosity and future depressive symtom atology once sociodemographic variables and covariates are controlled. With no associati on demonstrated between the two variables, a test for mediation could not be completed. Although the findings did reveal support for the hypothesis that religiosity is significantly associ ated with future lif e satisfaction, once sociodemographic variables and covariates were controlled, no evidence of a significant mediation effect of optimism was found. Predicting Depressive Symptomatology in Older Adults The data analyses provided several unexpected findings. As expected, bivariate analyses demonstrated that religiosity (i ndependent variable) wa s significantly, though weakly, negatively correlated with baseline depressive symptomatology ( r = -.10, p<.05). Unexpectedly, however, it was not significantly correlated with depressive symptomatology at Time 10 (dependent variab le), nor did it demonstrate a significant
170 association in regression anal yses predicting Time 10 depressive symptomatology. In this sample of older adults, there was no evidence that individuals with higher levels of religiosity scored significantly lower on de pressive symptomatology than individuals with lower levels of religiosity. This fi nding is somewhat inconsistent with numerous studies which have found various measures of religious involvement to be significantly associated with mental health outcomes such as depression and anxiety and other indicators of subjective well-being (e. g., Ellison, 1994, 1995, 1998; Fehring, Miller, & Shaw, 1997; Helm, Hays, Flint, Koenig, & Bl azer, 2000; Idler, 1987; Idler & Kasl, 1997; Jensen, Jensen, & Wiederhold, 1993; Kennedy, Kelman, Thomas, & Chen, 1996), Koenig, Hays, George, Blazer, Larson & Landerman, 1997; Levin, 1994; McCullough & Larson, 1999; Meisenhelder & Chandler, 2002). However, Koenig (2001) reported in a review of studies investigating the associa tion between religiousness/spirituality and mental heath that the findings have been mixed. Likewise, Hackney and Sanders (2003) produced meta-analysis results indicating that the relationship between religiousness and mental health outcomes varied as a function of the operationalizati on of the constructs being studied. The findings from this study appe ar to be more consistent with those of Wink, Dillon and Larsen (2005), who found religi ousness to be unrelated to depression in their sample of relatively healthy older adults. As stated previously, the conditions for me diation were not met in the analyses to predict Time 10 depressive symptomatology therefore optimism (mediator) could not be tested as a mediator. However, when optim ism was added to the model as an additional independent variable, it did demonstrate significant influence in the prediction of depressive symptomatology ( = -0.13, t = -2.11, p<.05), and its addition increased the
171 proportion of variance accounted for in the depend ent variable. In the final model, results indicated that lower levels of optimism were si gnificantly associated with higher levels of depressive symptomatology. These findings ar e consistent with a growing literature that has linked optimistic orientation with more optimal outcomes (e. g., Aspinwall & Taylor, 1992; Birren & Renner, 1980; Brenes, et al ., 2002; Fredrickson, 2001; Peterson, 2000; Scheier, et al., 1994). Previous researchers have also used the Li fe Orientation Test (LOT), to demonstrate that dispositional optimism remained a significant predictor of future optimism even after other possible overl apping variables such as neuroticism, trait anxiety, and mastery were controlled (e g., Scheier & Carver, 1987; 1994; Vickers & Vogeltanz, 2000). Vickers and Vogeltanz (2000) found that measures of dispositional optimism could be used to predict future depression over and above initial levels of depression, positive and negative affect, attributional style and daily hassles. Their work is perhaps most similar to this study in that they used a revised version of the LOT as was used in this study. They concluded that variables that predicted depression over the effects of Time 1 depressive symptoms, su ch as dispositional optimism, could be considered more unique predicto rs of future depression. Since it is generally accepted that dispositional optimism is stable over time, it is possible that its association with depressive symptomato logy represents a protective mechanism that serves to decrease an i ndividualÂ’s vulnerability to depressive symptomatology throughout the life course. Op timistic orientation may promote more adaptive coping capabilities that serve to bett er prepare individuals for the challenges and changes they will undoubtedly encounter in ol der age. Descriptive analyses indicated that mean levels of depressive symptoma tology for the final study sample did change
172 over the ten years between baseline measur ement (mean = 6.46, SD = 4.85) and Time 10 measurement (mean = 9.82, SD = 5.73). Thus, there was change over time in mean levels of depressive symptomatology for the sample, as would be expected with increasing age. With the current study it is not possible to determine if optimism might be exerting mediating influence on the rate of change in depressive symptomatology levels in the individuals in th e study. It could be argued that this particular sample might exhibit widespread optimistic orientation and reduced vulnerability to maladaptive changes associated with aging in general. Thus, they may be less vulnerable to depressive symptomatology. They appear to be at least somewhat financially secure, have chosen to relocate to a specially designed retirement community where they are surrounded by peers of similar backgrounds, ha ve considerable access to social support, recreational outlets, transporta tion, health care, and volunt eer opportunities. Almost 70% of the sample reported that they perceived thei r health to be good or very good, and they reported minimal disability. They have chosen to live in a community that encourages an active, engaged lifestyle. Thus, in light of the resources available to them, they may be less vulnerable to depression in general, or less likely to develop depressive symptoms over time, regardless of optimistic orientation. But why was Time 10 depressive sympto matology not predicted by religiosity in this sample of older adults? Part of the answer may be found in examining why higher levels of religiosity were e xpected to be associated with lower levels of depressive symptomatology. It has been suggested that religious involvement may impede depression in a number of ways, including, but certainly not limited to, providing social interaction, increasing per ceived support and a sense of belonging or embeddedness in
173 the community, enhancing self-esteem and sens e of personal mastery and sense of control over oneÂ’s personal affairs, fostering feeli ngs of worth and meaningfulness in life, promoting self-care and positive decision making (Beck, 1967; Cohen & Willis, 1985; Durkheim, 1901/1933; Ellison, 1991; 1994; Pearlin & Schooler, 1978; Wethington & Kessler, 1986). Perhaps, in this sample, many of the resources expected to be garnered from religiosity are also readily availabl e in the respondentsÂ’ environment. The respondents in this study reside in an e xpansive but tightly-knit condominium complex that affords them easy access to transportation, medical care, recreat ional resources, and daily social interaction with peers of thei r age group. Residents are encouraged to participate in community planning, decision making, and activities. Each building has a welcoming committee and a designated resident to check on shut-ins and represent the residents on community decisions. The reside nts have control over their own personal dwelling, while the complex staff attends to the faade as well as security, lighting, roadways, etc. Over 50% of the respondents reported that they did not live alone. And the majority reported membership in a local religious institution. T hus, for many of these older adults, religious involvement may actually provide redundant resources. An alternative explanation for the la ck of significant association between religiosity and depressive symptomatology mi ght be that the relationship is actually masked by the high level of religiosity in this sample, which made it difficult to find any religiosity effects. Or perhaps, it is a re sult of limited variability in the sample on the measures of depressive symptomatology. As discussed previously, this sample is not a precise representation of the general population, nor is it a clinical sample, and the level of depressive symptomatology was generally low, and there was not a great deal of
174 change in the mean level over the 10 year study period. This narrowed variability may have attenuated the relationships. In the current study, neither chronic h ealth conditions nor physical function and disability variables demonstrated significant predictive power of either Time 10 depressive symptomatology or Time 10 life satisf action. This, too, was surprising in light of research which has suggested associati ons between poor mental health outcomes and well-being, and declines in h ealth and function and chronic illnesses (e. g., Bisschop, et al., 2004; George & Fillenbaum, 1985; Larsen, 1978) often thought to be an unpreventable part of increasing age. Perhap s the lack of significant association between physical health and function and well-being outcomes found here is a product of the better-than-average health, education, and financial security characteristics of the sample. At baseline, sample respondents reported ve ry low levels of IADL and ADL difficulty, with slightly over 83% reporting no IADL di fficulties and almost 97% reporting no ADL difficulties. More than half of the sample reported 2 or fewer chronic conditions. Although purposely controlled to reduce the possi bility of spurious associations due to their influence, it was nevertheless anticipat ed that the health and function variables would provide significant explanatory power in the regression models. However, it is possible that their lack of significance in the model is a demonstration of the principle of diminishing returns Â– that because there were a moderate number of correlated predictor variables that accounted for meaningful amount s of unique variance, variables added late in the model-building process accounted fo r smaller and smaller proportions of unique variance (Hatcher & Stepanski, 1994).
175 An extremely large proportion of the samp le (88%) reported 12 years or more of education, with almost 61% indi cating that they had more than 12 years of education. All respondents indicated that they owned at least one home, and over half of them indicated that their total assets were above $175,000. There is copious literature from a number of research fields that demonstrates that edu cation and income, often combined to represent socioeconomic status, are strong predictors of more optimal outcomes. Thus, since this sample reports high levels of education and re latively high levels of financial security, high levels of physical and psychological func tioning would not be extraordinary (e. g., Berkman, et al., 1993; Guralnik, et al., 1993; Snowdon, Ostwald, & Kane, 1989). Consistent with those characteristics, an inverse relationship with depressive symptomatology was expected. Interestingly, education was not co rrelated with Time 10 depressive symptomatology, nor was it significant in the regression model. Likewise, total assets, though significantly correlated with Time 10 depressive symptomatology (r = -.12) was not a significant predictor in the regre ssion model predicting Time 10 depressive symptomatology. Since it has been repeatedly demonstrated that elderly adults with higher socioeconomic status tend to exhibi t more enhanced well-being and are more likely to be considered to be aging succe ssfully (Aneshensel, 1992; 2002; Baltes & Lang, 1997; Thoits & Hewitt, 2001; Wilson, 2000), it was expected that education and financial assets would demonstrate significance throughou t the analysis process. The lack of significant predictive power of these variables may be due to the lack of variance in the values, in that the data was somewhat skewed toward higher values.
176 With the exception of age, the remaining sociodemographic variables also failed to demonstrate significant influence on Time 10 depressive symptomatology in the final regression model. Although ther e are findings in the previous literature suggesting that depressive outcomes are associated with mar ital and living status (e. g., Berkman, et al., 1993; Bisschop, et al., 2004; Bosworth & Schaie, 1997; Kahana, et al., 2002; Ross & Mirowsky, 1989; Strawbridge et al., 1996), th e lack of significan t findings for these indicators in this study may be attributed to the fact that th ey were measured at baseline, whereas the depressive symptomatology measure that served as the criterion variable was measured 10 years later. It is quite possible that the respondentÂ’s ma rital or living status could have changed during the years leading up to the Time 10 measurement. It is therefore more likely that the criterion measurement reflects the influence of the respondentsÂ’ living status at Time 10 rather than at baseline. The importance of controlling negative affect when investigating the predictive ability of other variables in the study of depression has been suggested in the literature (Elliot, Marmarosh, & Pickelman, 1994). In this study, somewhat inconsistent with previous literature, negative affect, though highly correlated with Time 10 depressive symptomatology (r = .41, p<.05), was not sign ificant in the final regression model. Based on research that has demonstrated th at higher levels of negative affect are predictive of greater incidence of negativ e emotions, including depression (Chang & Sanna, 2001; Chang, et al., 1997; Fratiglioni et al., 2000; Kivela & Pahkala, 2001; Hillers, et al., 1998; Willis, 1986), it was e xpected that NA would have significant influence in models predicting depressive sy mptomatology. The fact that this did not prove to be the case is quite possibly a resu lt of redundancy nega tive affect actually
177 overlapped with the measure of Time 1 depr essive symptomatology. In cases where two measures are repetitive or redundant, the for both variables in the regression model are reduced since they are splitting their influence on the criterion variable (Pedhazar, 1997). Predictors of Depressive Symptomatology The indicators that demonstrated significant predictive power in the prediction of Time 10 depressive symptomatology were age, health self-appraisal, positive affect and baseline depressive symptomatology. The most powerful predictor of Time 10 depressive symptomatology in this sample of older adul ts was the level of health self-appraisal reported at baseline ( = -0.21, t = -3.34, p<.05). There was an inverse relationship between the two measures, where the higher th e respondentÂ’s rating of health status at baseline, the lower the level of depressive symp tomatology at Time 10. This is consistent with much previous research that has indicate d that self-reported hea lth status, even when posed as a single question, is a powerful pred ictor of a wide range of outcomes. The findings here make sense intuitively, since it is more likely that an individual will report less psychological distress if they are experiencing higher levels of perceived health. It is also possible that individuals who report higher levels of se lf-rated health are more likely to be positive in other aspect s of their life as well and, thus, report fewer symptoms of psychological distress. Further, one cannot rule out the possib ility that measures of selfrated health are influenced by social compar ison bias or social desirability, whereby respondentsÂ’ perception of their health status is influenced by a comp arison to the health status of those around them (social comparison) or a desire to create a certain image of
178 themselves as either more healthy or less he althy (desirability bias) with their choice of response. Although not the most powerful predictor, baseline depressive symptomatology was found to be significant in the predicti on of Time 10 depressive symptomatology. This is somewhat, though not totally, consiste nt with previous research which has found baseline depressive symptoms to be the most powerful predictor of future depressive symptoms (Lewinsohn, Roberts, Seeley, Rohde, Gotlib, & Hops, 1994; Skevington, 1993; Vickers & Vogeltanz, 2000). Since the relationship between low positive affect and depressive symptoms has been well established in the literature (Jolly, Dyck, Kr amer, & Wherry, 1994; Watson, Clark, & Carey, 1988; Vickers & Vogeltanz, 200 0), it was expected that baseline positive affect measures would be associated with Ti me 10 depressive symptomatology. This was indeed the case, with positive affect dem onstrating a negative relationship with the criterion variable such that lower levels of positive affect were associated with higher levels of depressive symptomatology. This appears to support previous work suggesting that mood serves as an important factor in psychological adjustment (Chang & Sanna, 2001) and well-being. High positive affect is characterized by high energy and pleasurable engagement and has been linked to positive physiological changes (Futterman, et al., 1994), better social relationships (Taylor & Brown, 1988), maintenance of conscientious health habits, and appropriate use of healthcare services (Taylor, et al., 2000). In turn, each of these variables would seem to serve protective roles in reducing the likelihood of increased depr essive symptoms over time. Whereas, low positive affect is characterized by sadness a nd lethargy and is inversely correlated with
179 physical disability, cognitive problems, negative life events, and minimal social contacts (Pitkala, et al., 2004). As expected, age was also a significant predictor of Time 10 depressive symptomatology, indicating that increasing age was significantly associated with increases in levels of depressive symptomatol ogy, which is consistent with the literature. Increasing age is also associated with declin es in physical health and function and, in general, with survival and quality of life i ssues (e. g., Laukkanen, et al., 2000; Scott, et al., 1997 Vaillant, 1977). In turn, age-associated increases in chroni c illness and negative self-appraisals of health are predictive of lowe r levels of morale and a general absence of well-being (Bisschop, et al., 2004; Larsen, 1978). However, in this sample, age proved to be the least influential of the significant indicators of depressive symptomatology. Again, the timing of the data collection may pl ay a role in this finding. Since age at baseline was the measure used as a control and the dependent variable was measured 10 years later, it is possible that the true association between the two measures was under estimated. Descriptive analyses for the original study sample at baseline, at Time 5 and at Time 10 did indicate that mean levels of depressive symptomatology for the sample increased over time. It is also possible that th is sample is not representative of an average population of older adults. As stated previ ously, the sample in this study was somewhat more advantaged in terms of education, financial security, and health status. Additionally, as described previously, the age-segregated environment in which they reside may serve to buffer some of the deleteri ous effects associated with old age as well. These resources may serve to assist these aging adults in negotiating challenges often
180 associated with depression in the latter stages of the life course, such as social isolation, poor health care, reduced mobility and lack of recreational outlets. Predicting Life Satisfaction The guiding hypotheses for this study also postulated that, onc e control variables and covariates were held constant, religiosity would be positively associated with Time 10 life satisfaction (hypothesis 1.1), and that the relations hip between the two would be mediated by optimism (hypothesis 2.1). Results from the regression analyses performed here support only the first hypothesis. Relig iosity demonstrated significant, though less than expected, influence in the prediction of future life satisfaction. Results indicated that individuals who reported greater levels of religiosity at baseline were more likely to report greater levels of life satisfaction at Time 10 ( = 0.15, t = 2.81, p<.05). These findings are consistent with previous studi es demonstrating a si gnificant association between religious involvement and life satisfaction (e. g., Armstrong & Goldsteen, 1990; Cox & Hammonds, 1988; Cutler, 1976; Ellis on, 1991; Hodges, 2002; Hunsberger, 1985; Koenig, George & Siegler, 1988; Kvale & Fe rrel, 1988; Levin & Taylor, 1998; Morse & Wisocki, 1988; Neill & Kahn, 1999). The findings are also somewhat similar to those of Morris (1991), who found in a study with olde r adults that life satisfaction was best predicted by subjective health status, income satisfaction and church attendance. In the current study, health self-appraisal was also significantly correlated with Time 10 life satisfaction (r = .23, p<.05) as well as signifi cant in each of the hierarchical regression models. However, total assets reported at baseline, representing financial status of respondents in this sample, though significantl y correlated with Time 10 life satisfaction
181 (r = .15, p<.05), were not significant in a ny of the regression analyses. Since the literature generally suggests that income is strongly associated with life satisfaction, it was thought that financial assets would be significantly infl uential in predicting future life satisfaction. Given the current trend toward great increases in cost of living expenses, this could have important implications for current and future retirees, particularly those who are faced with limited financial resour ces. These findings seem to suggest that financial security, in and of itself, may not n ecessarily predict well-being in later years. How might religiosity influence life satisfaction? Many explanations have been suggested. For example, religious involve ment may promote or facilitate social engagement, may provide a structure to help individuals find meaning and coherence in life, may facilitate the creation of a persona l relationship with a higher power, and may assist individuals in coping with traumatic events, losses or life changing experiences. Perhaps, for some individuals, it is the stru cture of church attendance Â– a reason to dress up, to leave the house, and to engage with peers. For older people in particular, religious involvement may provide a place to go, a place to connect; it may provide familiar, positive memories (e. g., weddings, baptisms). For some, the church may provide a sense of belonging, a sense of bei ng a part of a community of like-minded individuals that make up the congregation. For older adults fa cing the challenges of aging, religious involvement may provide a sense of stabilit y Â– something that they can count on to remain consistent and welcoming Â– even as life changes around them. Religious teachings and beliefs may enhance resilien cy by providing comfort and hope at a time when older adults are facing dec line and deficits inherent in th e aging process, or as they are confronted with the realities of their own mortality. They may help to define meaning
182 and purpose in life despit e age-related declines, hardship or adversity and enable one to persevere in the face of challenges. Religious involvement has also been linked to a broad array of positive health practices and psychosocial benefits (Idler & Kasl, 1997). Religiosity may enhance an individualÂ’s desi re to practice healthy behaviors and their belief that they have responsibility for, a nd control over, their ow n healthy choices and behaviors. These factors may contribute to an in creased sense of satisfaction in oneÂ’s life. The most salient predictor of future life satisfaction in this study was life satisfaction reported at baseline ( = 0.30, t = 5.03, p<. 05). Although there was a decrease in the overall mean life satisfaction levels reported for the sample at Time 1 (mean = 19.11, SD = 3.39) and Time 10 (mean = 18.65, SD = 3.63), respondents who reported higher life satisfacti on at Time 1 were significantly more likely to report higher life satisfaction at Time 10. This is consistent with the generally accepted belief that life satisfaction remains relatively stable across the life span. The data also revealed a number of unexpected nonsignificant findings. For example, marital status was not significant in any of the regression models and, once religiosity was added, living status was no l onger significant. It had been expected that older adults who were married or not living alone would be significantly more likely to find satisfaction in life. It was thought that living with someone or being married would provide a number of advantages that ma y enhance subjective well-being, such as, providing ready access to a support system, having someone to confide in and count on in stressful or challenging times, providing an increased sense of being needed or appreciated, loved, or cared for and the opportu nity to reciprocate, having a partner with whom to participate in activities, and cont ributing to a reduced sense of loneliness.
183 However, the findings here suggest that the institution of marria ge, often considered protective, may not provide the sa me level of advantage, or remain as influential as adults reach increasingly greater ages. Perhaps fo r this sample of older adults, living arrangement, married or not, takes on less importance due to the ready availability of support and the closeness of neighbors and fr iends in their own retirement complex. Perhaps social supports and a close-knit comm unity, such as this retirement community, provide similar protective effects as have been associated with the institution of marriage. It has been suggested that the highest le vels of well-being are found among people who believe they have high levels of social support available to them (Ka hn, et al., 2003). This certainly has important implications in a population that will see increasing numbers of elderly adults making retirement-living choi ces in the coming decades. Living alone may also force an individual to become self-su fficient and to develop other support systems, such as religious involvement, to help to meet needs otherwise associated with marriage. It might be that living alone despite the chal lenges associated with increasing age, or the ability to maintain independence after the loss of a spouse serves as somewhat of a Â“badge of honor,Â” representing the elderly individualÂ’s resiliency and ability to successfully cope with adversity. This, in turn, may enhance self-esteem and a sense of mastery and lead to greater life satisfaction. Living alone also gives an individual more control over their interactions with others. For example, by allowing them greater choice over who they will interact with, and when, it may be possible to reduce stress associated with interpersonal conflict or negative intera ctions. Individuals who live alone also may not experience the stress that may result from havi ng to care for an ill spouse or having to give up activities or plans in or der to meet a spouseÂ’s needs.
184 Another surprising finding was the lack of significant associa tion between health and function variables and futu re life satisfaction. The liter ature has suggested that poor health and level of difficulty with Activities of Daily Living (ADLs) are associated with life satisfaction (e.g,.Meeks & Murrell, 2001) Yet, although chronic conditions, IADLs, and ADLs were significantly correlated with Time 10 life satisfaction, none of these constructs demonstrated significance in th e regression models. Again, this may be a reflection of the low levels of disability a nd high levels of health reported by the study participants. It seems likely that in this healthy sample, life satisfaction levels might be inflated due to the lack of effects from fact ors that are frequently associated with lower levels of well-being and satisfaction w ith life in the gene ral population. Also unexpected was the lack of signif icant association of the mood constructs and Time 10 life satisfaction even though bot h NA and PA were significantly correlated with it. Negative affect (NA) did not reach significance in any of the regression models. Whereas positive affect (PA) was significant in the first model ( = 0.13, t = 2.24, p<.05), but failed to remain significan t once religiosity was added to the model. This may be indicative of an overlap be tween the two constructs since empirical research has demonstrated that religion is associated with a generalized expectation that outcomes will be positive (Carver, et al., 1989; Idler & Ka sl, 1992). Perhaps they are actually sharing variance in the prediction of life satisfaction, although diagno stic tests did not reveal multicollinearity issues. It is more likely an indication that religiosity is correlated with positive mood states but offers benefits above and beyond the effect of PA. It may also be another reflection of the selectivity of this sample. In general the sample reported relatively good health, moderately high levels of PA and optimism and low levels of NA.
185 Thus, it may be that the lack of associati on between mood constructs and life satisfaction was a reflection of the limited variance in the measures which resulted in somewhat of a bias toward more positive outcomes. Optimism as a Mediator in the Relationshi p Between Religiosity and Life Satisfaction What is clear from these findings is that individuals who report higher levels of life satisfaction, religiosity, and positive health self-appraisal are more likely to report greater life satisfaction in the future. In a ddition, it has been sugge sted that religiousness has an effect on life satisf action through the influence of optimism (Salsman, Brown, Brechting, & Carlson, 2005). Regression anal yses in this study did not support those findings. Although religiosity and optimism were correlated in bivariate analyses, they were not significantly associated in multivar iate models that included the controls and covariates. Thus, causal steps mediational anal yses were not appropriate. However, when optimism was added to the multivariate model that included religiosity, it did account for a modest (5%), but significant increase (F[1 238] = 19.37, p<.001) in the proportion of variance in predicting life satisfaction above and beyond the variance accounted for by religiosity, demographic variables and cova riates. A Sobel Test to determine the significance of indirect effects exerted by optimism in the model failed to reach significance. Thus, contrary to expectations, these data suggest that higher levels of religiosity at Time 1 are not necessarily a ssociated with higher levels of optimism at Time 5, but optimism is significantly associated with future levels of life satisfaction.. This suggests that individuals who are more religious, or who tend to hold a more positive, favorable outlook on life, are more likely to report greater levels of life
186 satisfaction. The results do not, however, suppor t the proposition that optimism serves as one of the mechanisms through which reli giosity exerts its positive influence on subjective well-being. And, since religiosity was measured with a multi-item index, it is not clear if specific dimensions analyzed i ndividually may have produced the expected associations. As noted previously, the manner in which the construc ts are operationalized and measured is likely to affect the nature of the relationship between these measures. Further, since optimism was not measured at baseline in the original study, it could not be statistically controlled in these analyses. Fo r example, it might be that religiosity, measured at Time 5, could mediate optimism th at had been measured at baseline. Thus, many questions remain unanswered. For example, it is impossible to discern if the data are really reflecting the fact that more optimistic people tend to be more religious. Perhaps the more optimistic one is, the more lik ely it is that they will participate in or turn to religion as an adjustment resource. Or perhaps the more religious one is, the more optimistic they become over time, or the more likely they are to hold a world view that Â“the glass is half full versus half empty.Â” Although it has been ge nerally accepted in the literature that optimism, as a trait, remains somewhat stable across the lifespan (Shnek, Irvine, Stewart, & Abbey, 2001; Steel, & Wade, 2004), some researchers have argued that optimism is unlikely to entirely account fo r the effects of religiousness (Ai, Peterson, Bolling, & Koenig, 2002). For example, as a personality trait, optimism may exert its influence prior to the development of religious orientation or behavior Regardless, in this sample of elderly adults, it does not appear that optimism serves as a mechanism through which religiosity may influence life satisfaction.
187 Resilience and Elderly Adults It has been suggested that the most optimal outcomes in aging are achieved through a combination of protective factor s working together (e. g., Bergeman & Wallace; Rutter, 1987). An alternative interpre tation might be that the most resilient individuals are those individuals who have a combination of resources that work together to enable them to adapt to change, cope with declines and deficits inherent in the aging process, and rebound or rise above adversity. The results presented here are consistent with this line of reasoning. The fact that the respondents in this study have survived to join the ranks of the old or oldest-old adults could be viewed as a testament to their resilience. Despite their advanced age, participants reported relatively low levels of health and functional difficulties as well as high levels subjective well-being, as evidenced by their reported levels of life satisfaction and low levels of depressive symptomatology. Thus, not only are they survivi ng into advanced old age, they appear to be thriving. The findings suggest that there are a number of fact ors that contribute to their resiliency. Specifically, high levels of se lf-appraised health, positive mood states (positive affect), thei r interpretation of global levels of depressive symptomatology and life satisfaction, religiosity and optimism. Thus, these findings are consistent with previous works that have suggested that re silient individuals possess some combination of internal (e.g., personality or dispositional characteristics) and external resources (e. g., social engagement, prosocial behaviors such as religious involvement) that serve as protective factors that enhan ce their ability to adapt more successfully to the aging process.
188 Limitations of the Study Although the findings of this study provide further insight into how factors may work together as resilience resources to promote greater levels of subjective well-being in older adults, there are a number of limitations to consider when interpreting the findings and drawing conclusions. The external validity of this study was limited by the lack of racial and economic diversity of the part icipants. Although this sample of White, moderately healthy, upper middle-class, re tirement community-dwelling older adults was representative of the commun ity-dwelling elderly in the ar ea, the results may not be generalizable to all older populations (e.g ., lower socioeconomic levels, minorities, urban-dwelling, residents who do not live in ag e-segregated areas) or to more infirm or institutionalized populations (e. g., chroni cally ill, frail, disabled, nursing home residents). For example, previous research has indicated that religious involvement may be more important to African Americans th an Whites (Ellison, 1995), and that minorities, in general, are more likely to be more reli gious (Ferraro & Koch, 1994; Levin, Taylor, & Chatters, 1994). There are also strong indications in the literature that there are likely to be differences in outcomes based upon well-esta blished associations between such factors as income, education, functional status and depressive symptomatology. It may also be that the factors studied here may affect younger populations differently (i.e. the young old or old versus the old-old as studied in this sample). Thus, future research is required in order to determine if the results identified here are consistent with those of a more diverse group of study subjects. The sample in the current study was also limited due to the effects of both missing data and attrition. Respondents who were ex cluded due to missing data (N = 202 of the
189 original sample of 1000) were more likely to have reported lower levels of positive affect (PA), lower baseline life satis faction and greater levels of IADL difficulties. Thus, those respondents included in the sample were more likely to have a more positive mood and outlook on life and fewer physical disabilities than those who were excluded from the sample. This could have resulted in an inflati on of the scores for overa ll life satisfaction. The respondents lost to attriti on reported significantly lower self-appraisals of health, PA, and life satisfaction, and higher levels of both IADL and ADL difficulties. Thus, respondents who were still in the study after 10 years demons trated fewer risk factors associated with increased depressive symp tomatology and reduced life satisfaction than those who were excluded at baseline. Future work that includes mo re diverse populations could greatly improve the external validity of findings. Further, the CES-D utilized to measure depressive symptomatology in this study is a non-clinical, non-diagnostic tool that relies on self-report to screen for the incidence of depressive symptomatology. Clinical assessment by trained providers or an additional measurement provided by a key observer (e. g., spouse; peer) to corrobor ate the patient data would strengthen the findings. Another important consideration is the fact that the index used to measure religiosity was developed by the researcher with a series of items fr om the original study questionnaire. Although the index demonstrated moderate reliability (.87) (Table 1) in this study, it had not previously been validated One might also argue that the index was heavily focused on religious coping and may ha ve provided a less than ideal range of factors important to the study of religiosity. Use of a previously validated, widely used
190 scale to measure religiosity w ould greatly improve the genera lizeability of the results and would allow for replication in future studies. Another limitation is the self -report nature of the data used in this study. Self report may be especially problematic when re porting chronic illness, physical health and functional difficulties. Self-reports of h ealth are the products of the individualÂ’s perception and may reflect the re spondentÂ’s understanding of an il lness or disability or be tempered by their perception of the severity of the illness or infirmity and may not necessarily provide an accura te, objective measure of health or disability. Self-reports may also be vulnerable to social desirability or social comparison bias. Further, the use of dichotomous categories to measure chronic conditions is limiting since it does not account for the severity or duration of the i llness or difficulty (Mill s, 2001). It is likely that significant association between hea lth and function measures and the outcome variables in this study would not have been lacking if severity measures had been included in the analyses. Another limitation that shoul d be considered is the time between data points. The ten year spread between baseline measures and outcome measures may have led to greater levels of attrition and an under-es timation of the associations between the variables. Health and function declines and chronic illness are an inherent part of the aging process. With increasi ng age, there is often an in creased likelihood of onset of health issues that may impact an individua lÂ’s perception of the qua lity and meaning of life, their expectations of the future, their desi re and/or ability to so cialize or participate in important activities, their mood states and thei r general sense of well-being. It is likely that measures for the covariates taken at baseline would have presented stronger
191 associations with depression a nd life satisfaction as the leve ls of decline increased with age. A further limitation in this study was imposed by the use of secondary data. When using secondary data for analyses, the researcher may be constrained by the data that are available and frequently must make ad justments in the conceptual model. In this study, it would have been preferable to have a measure of optimism at Time 1 in order to test for possible mediation effects of religiosity on optimism. Also, in order to increase the comprehensiveness of the religiosity index, measures of spirituality, prayer habits, past religious involvement or behaviors, a nd further indicators of religious salience, would have been pertinent additions if they ha d been available in the waves of data chose for analysis. Strengths of the Study Since this study utilized longitudinal data with multiple data points it was possible to control a number of important sociodemogr aphic variables, initial measures of the dependent variables as well as a number of covariates that have previously been identified as possible confounders, such as positive and negative affect, health selfappraisal, chronic illness and IADL and ADL difficulties. For example, negative affect has been found to overlap with and e xhibit strong influence on depressive symptomatology (Kahn, et al., 2003; Watson & Clark, 1984). Health, functional status, and mobility difficulties can undermine the asso ciation between religi ous involvement (e. g., attendance; activities), subjective health appraisals and optimism (Musick, 1996; Oman & Reed, 1998). Control of confounders and covariates allows for a better
192 understanding of the associations between study variables of in terest and further explanations of the streng th of their influence. An additional strength for this study lies in the operationalization and measurement method used for the independent variable. Many earlier studies have used religious attendance as a proxy measure to repr esent the construct of religiosity. In the case of elderly adults, results might be sk ewed since the relationship between age and religious attendance is generally thought to be curvilinear Â– as age goes up, attendance often goes down (e.g., functional disability reduces mobility). Therefore, religiosity should be treated as a complex concept that most certainly represents more than a measure of attendance at religious services. Based upon the data that were available in the original study, a index was constructed in order to create a more comprehensive measure of religiosity. Indicators reflecting organizational religios ity, non-organizational religiosity, subjective religiosity (salience) and religious support and comfort (coping) were combined in an effort to capture an array of elements thought to be important in defining and measuring religiosity. Implications The findings of this study contribute to the extant literat ure demonstrating protective effects of religiosity and optimis m for aging adults. Specifically, it has been shown that higher levels of religiosity a nd higher levels of optimism in retirementcommunity-dwelling older adults are predictive of greater future life satisfaction. Life satisfaction, in turn, has long been suggested as the most salient indicator of successful aging (Havinghurst, 1961; Rudinger, & Thomae, 1990).
193 Powell and colleagues (2003) have suggested that, in addition to providing religious services, religious congregations are readily avai lable community institutions that may offer access to a broa d array of inexpensive resource s. Encouraging older adults to engage not just in religious services, but to participate in outr each programs and other activities offered by community congregations may be cost-effective ways of channeling resources to those who need assistance in m eeting some of the challenges of aging. Congregational outreach services, such as in-home visitations, prayer lines, phone support for shut-ins, or parish nurse services, may be especially important for older adults who are challenged by infirmities and mobility difficulties (e. g., no longer able to drive) that impact their ability to attend services or access other church activities. Widely available, easily accessible educational opport unities are needed in order to enlighten community service providers, physicians, public health work ers, and other providers on the links between religiosity and well-b eing and to encourage networking amongst potential resources. Community leaders a nd service providers need to develop an understanding of such concepts as religiosi ty and spirituality, and determine how they might be used effectively in policy and practice (Davis, 2005). Koenig (2004) has suggested that academic preparation, traini ng, attending continuing education courses and reading the related literature are all ways of becoming informed about the religion-wellbeing connection. Service provi ders should assess their pati entsÂ’ / clientsÂ’ level of religious or spiritual belief and functioning and its impact on their well-being (Emmons, 2005) and be cognizant of their importance in the lives of many individuals. Those who are not comfortable with these concepts should be prepared to make appropriate referrals in order to better serve their clients a nd avoid undermining potential resources.
194 In recent years research ers focused on successful agi ng have begun to investigate factors that protect older adults from risk situations or a dverse outcomes. Some of this research has suggested that there are likely late nt resources that play a significant role in helping older adults remain resilient despite th e changes and declines that are inherent in the aging process (e. g., Staudinger, et al., 1993). The results of th e research presented here indicate that dispositional optimism may be one such resource. For example, assessment of an individualÂ’s level of di spositional optimism may provide a means for identifying individuals who may lack an im portant resource to successfully adapt to challenges posed by the aging process, illness, or declines in health and functional ability. Assessment of optimistic orientation and hope may provide an important means of identifying individuals who are less likely to be resilient in the face of adversity or are at greater risk for adverse outcomes. The results here indicate that individuals with higher levels of optimism are more likely to report hi gher levels of future subjective well-being, which previous literature has shown to be a significant predictor of successful aging. Thus, perhaps Watty PiperÂ’s (1930) little blue engine already knew what researchers are now beginning to discover, that Â“ I think I can I think I can, I think I canÂ” may be more than a childÂ’s storybook lesson. It may repr esent a life-long perspective that has a powerful influence on the pos itive aging experience of older adults. Suggestions for Future Research There has been increasing r ecognition that religiosity is associated with a number of positive physical and mental health and well-being outcomes. Consistent with previous studies, these findings suggest that religiosity is an important dimension in the lives of older adults. Growing evidence suggest s that religiosity serves as an important
195 resiliency resource. With this has come the development of a vast array of conceptualizations, definitions and measures of religious involvement and spirituality. However, it is becoming clear that more work is needed to develop religion and spirituality measures that are theoretically and functionally linked to important facets of physical and mental health and are consiste ntly employed in research. The literature clearly suggests that the nature of the re lationship between religiosity measures and physical and mental health outcomes varies by population and by the manner in which these constructs are operationa lized and assessed. More wo rk is needed in order to determine differences and similarities between the constructs of religiosity and spirituality and their respective effects on a wide range of outcomes. Future studies that create and utilize standardized measuremen t and analytical techniques with a more heterogeneous sample are needed. In particular, longitudinal studies that help to elucidate the temporal manner in which these constructs relate to one anothe r would greatly expand our understanding. Research is needed to examine the potential age differences in the effects of religiosity and th e relative importance of each of the possible domains of influence. Further research is still required to uncove r the mechanisms through which religiosity exerts its influence and what fact ors may mediate or moderate its effects. Especially pertinent to the aging population w ill be research that focuses on more solitary measures of religiosity, such as prayer and spirituality, which may increase in importance as mobility and social engagement become more restricted with advancing age. Given the increasing older population and concerns rela ted to overburdening of health care and related services, subsequent research should explore the re lationship between religiosity and other factors that may serve as resilien cy resources for older adults. For example,
196 links between religiosity and social support, volunteerism, meaning and purpose in life, and various aspects of personality and self -concept such as, personal mastery, coping styles, self-esteem, and altruist ic orientation. Future stud ies are needed to explore the vast array of potential mediator or moderato r variables and identify which constructs are particularly useful in further articulati ng the relationship between religiosity and subjective well-being. Further, it will be espe cially important to consider these factors in conjunction with the various domains of relig ious influence as it is likely that as individuals age the importance of specific elem ents of religiosity will vary or change. Extant literature on optimism clearly indi cates that optimism predicts a wide array of positive outcomes in such areas as physic al health and psychological well-being. The findings of this study add to the growing ev idence of the importance of dispositional optimism as a resiliency resource in older adu lts. However, further research is needed in order to disentangle the possible overlap wi th other personality constructs such as neuroticism and extraversion. Longitudinal studi es are needed in orde r to articulate the relationship between these two constructs over the lifespan and to explore the feasibility of interventions to positively impact levels of dispositional optimism. Future work should also include the exploration of the developmen tal origins of optimism or the determinants of optimism among older adults. If disposit ional optimism is indeed stable across the lifespan, what strategies can be utilized to he lp sustain or enhance this orientation? Is optimism a truly a linear variable ? Are overly optimistic indivi duals able to sustain their optimism over time, or are th ey less likely to have positive outcomes in the future? Gaining an understanding of how some indivi duals are able to maintain a sense of optimism in the face of life long challenges, the development of assessment tools for
197 identifying those adults who are at greatest risk of reporting lo w levels of life satisfaction, and the creation of interventions that can a ssist older adults in gaining or sustaining optimistic evaluations of their aging experi ence should all be goals for the future. In the future, researchers should also c ontinue to search for evidence as to the most valid manner in which to operationaliz e optimism Â– as a unidimensional model in which optimism and pessimism are polar opposites, or as a two-dimensional model in which optimism and pessimism are treated as tw o separate constructs Further, although assessment of optimism can be done simply and with minimal cost or resources (e. g., the LOT is available in the public domain and does not require specialized scoring procedures), the development of alternative a ssessment strategies that do not rely entirely on self-report would be useful in establishing convergent validity.
198 References Ai, A. L., Peterson, C., Bolling, S. F., & Koenig, H. (2002). Private prayer and optimism in middle-aged and older patients awaiting cardiac surgery. The Gerontologist, 42 (1), 70-81. Ainlay, S. C. & Smith, D. R. (1984). Aging and religious participants. Journal of Gerontology 39 357-363. Ainlay, S. C., Royce, S. Jr., & Swigert, V. L. (1992). Aging and religious participation: Reconsidering the effects of health. Journal for the Scientific Study of Religion, 31 175-188. Albert, S. M., Im, A., & Raveis, V.H. (2002). Public Health and the second 50 years of life. American Journal of Public Health 92 (8), 1214-1216. Aldwin, C. M., Levenson, M. R. & Spiro, A. III. (1994). Vulnerability and resilience to combat exposure: Ca n stress have li felong effects? Psychology and Aging, 9 (1), 34-44. Allison, P. D. (1999). Logistic regression using the SAS System. Theory and Application Cary, NC: SAS Institute Inc. Andrews, G., Clark, M., & Luszcz, M. (2002). Successful aging in the Australian Londitudinal Study of Aging: Applying the MacArthur Model cross-nationally. Journal of Social Issues, 58 (4), 749-765.
199 Andrews, F. M., & Withey, S. B. (1976). Social indicators of well-being New York: Plenum Press. Aneshensel, C. S. (1992). Social stress: Th eory and research. In J. Blake, & J. Hagan (Ed.), Annual review of sociology Palo Alto, CA: Annual Reviews. Aneshensel, C. S. (2002). Theory-based data analysis for the social sciences Thousand Oaks, CA: Pine Forge Press. Antoni, M. H. & Goodkin, K. (1988). Ho st moderator variables in the promotion of cervical neoplasia-I. Personal facets. Journal of Psychosomatic Research, 32 (327-338. Aquino, J.A., Russell, D.W., Cutrona, C.E., & Altmaier, E.M. (1996). Employment status, social support, a nd life satisfaction among the elderly. Journal of Counseling Psychology, 43 (4), 480-489. Armstron, M. J., & Goldsteen, K. S. (1990). Friendship patterns of older American women. Journal of Aging Studies, 4 391-404. Aspinwall, L. G., & Taylor, S. E. ( 1992). Modeling cognitive adaptation: A longitudinal investigation of the impact of individual differences and coping on college adjustment and performance. Journal of Personality and Social Psychology, 63 9891003. Atienza, F. L., Balaguer, I., & Garcia-Mer ita, M. L. (2003). Satisfaction with Life Scale: Analysis of fact orial invariance across sexes. Personality and Individual Differences, 35 1255-1260. Bachay, J. B., & Cingel, P. A. (1999) Restructuring resilience: Emerging voices. Affilia, 14 (2), 162-175.
200 Baltes, P. B., & Baltes, M.M. (1990). Psychological perspectives on successful aging: The model of selectiv e optimization with compensation. In P.B. Baltes & M..M. Baltes (Eds.), Successful Aging: Perspectives From the Behavioral Sciences (pp. 1-34). New York: Cambridge University Press. Baltes, P. B. & Baltes, M. M. (1993). Successful aging: Perspectives From the Behavioral Sciences Cambridge, MA: Cambridge University Press, European Science Foundation. Baltes, M. M., & Carstensen, L. L. (1996). The process of successful ageing. Ageing and Society, 16 397-422. Baltes, M. M., & Lang, F. R. (1997). Ev eryday functioning and successful aging: The impact of resources. Psychology and Aging, 12 (3), 433-443. Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research : Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 52 1173-1182. Bauman, S., Adams, J. H., & Waldo, M. (2001). Resilience in the oldest-old. Counseling and Human Development, 34 (2), 1-20. Bearon, L.B. (1996). Successful aging: What does the Â“good lifeÂ” look like? The Forum For Family & Consumer Issues. [Online]. Available: http//www.ces.ncsu.edu/edu/depts/fcs/pub/ aging.html. Accessed: 5/01/00. Beck, A. T. (1967). Depression: Causes and treatment Philadelphia: University of Pennsylvania Press.
201 Beckett, L. A., Brock, D. B., Lemke, J. H., Mendes, d. L. C., Guralnik, J. M., Fillenbaum, G. G., et al. (1996). Analysis of change in self-re ported physical function among older persons in four population studies. American Journal of Epidemiology, 143 776-778. Bergeman, C. S., & Wallace, K. A. (1999). Resiliency in later life. In T.W. Whitman, T. V. Merluzzi, & R. White (Eds.), Psychology and medicine: The Life-span developmental perspective (pp. 207-225). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Berkman, L. F., Seeman, T. E., Albert, M ., Blazer, D., Kahn, R., et al. (1993). High, usual and impaired functioning in community-dwelling older men and women: Findings from the MacArthur Foundation Research Network on Successful Aging. Journal of Clinical Epidemiology, 46 1129-1140. Berkman, L. F., & Syme, L. (1979). Social networks, host resistance, and mortality: A nine-year follow-up st udy of Alameda County residents. American Journal of Epidemiology, 109 (2), 186-204. Binstock, R. H. (1997). Issues of resource allocation in an aging society. In: T. Hickey, M. A. Speers, & T. R. Prohaska (Eds.), Public Health and Aging (p.55). Baltimore: The Johns Hopkins University Press. Bisschop, M. I., Kriegsman, D. M. W., Beekman, A. T. F., & Deeg, D. J. H. (2004). Chronic diseases and depression: Th e modifying role of psychosocial resources. Social Science & Medicine, 59 721-733. Blazer, D. G. (1982). Social support a nd mortality in an elderly community population. American Journal of Epidemiology, 115 (5), 684-694.
202 Bloom, J. R. (1990). The relations hip of social support and health. Social Science and Medicine, 30 635-637. Board of Trustees, Federal Hospital In surance Trust Fund. (2000). 2000 annual report of the Board of Trustees of the Federal Hospital Insurance Trust Fund. Washington, DC: U. S. Government Printing Office. Boey, K. W. (1999). Cross-validation of a short form of the CES-D in Chinese elderly. International Journal of Geriatric Psychiatry, 14 (8), 608-617. Bosworth, H. B., & Schaie, K. W. (1997). The relationship of social environment, social networks, and health outcomes in the Seattle Longitudinal Study: Two analytical approaches. Journal of Gerontology: Psychological Sciences, 52B (5), P197-P205. Bradburn, N. M. (1969). The Structure of Psychological Well-being Chicago: Aldine. Brandstdter, J. & Greve, W. (1994). Expl aining the resilience of the aging self: Reply to Carstensen and Freund. Developmental Review, 14 93-102. Brandstdter, J. & Greve, W. (1994). Th e aging self: Stabilizing and protective processes. Developmental Review, 14 52-80. Brandstdter, J., & Renner, G. (1990). Tenacious goal pursuit and flexible goal adjustment: Explication of age-related analysis of assimilative and accommodative strategies of coping. Psychology and Aging, 5 58-67. Brandstdter, J., & Rothermund, K. ( 2000). The life-course dynamics of goal pursuit and goal adjustment: A two-process framework. Developmental Review, 22 117150.
203 Brandstdter, J., Wentura, D., & Greve. (1993). Adaptive resources of the aging self: Outlines of an emergent perspective. International Journal of Behavioral Development, 16 323-349. Brenes, G. A., Rapp, S. R., Rejeski, W. J., & Miller, M. E. (2002). Do optimism and pessimism predict physical functioning?. Journal of Behavioral Medicine, 25 (3), 219-231. Breslow, L., Reuben, D., & Wallace, S. (2000). Introduction to special issue on health promotion in the elderly. American Journal of Health Promotion, 14 341-342. Bromberger, J. T., & Matthews, K. A. (1996). A longitudinal study of the effects of pessimism, trait anxiety, and life stre ss on depressive symptoms in middle-aged women. Psychology and Aging, 11 207-213. Bruce, M. L., Seeman, T. E., Merrill, S. S., & Blazer, D. G. (1994). The impact of depressive symptomatology on physical di sability: MacArthur Studies of Successful Aging. American Journal of Public Health. 84 1796-1799. Buchner, D., Beresford, S., Larson, E ., LaCroix, A., & Wagner, E. (1992). Effects of physical activity on health status in older adults : II. Interventions studies. Annual Review of Public Health, 13 469-488. Bull, C. N., & Aucoin, J. (1975). Vo luntary association participation and life satisfaction: A replication note. Journal of Gerontology, 30 73-76. Campion, E. W. (1998). Aging better [Editorial]. The New England Journal of Medicine, 338 (15), 1064-1066.
204 Carver, C. S., Pozo, C., Harris, S. D., No riega, V., Scheier, M. F., Robinson, D. S., et. al. (1993). How coping mediates the effects of optimism on distress: A study of women with early stage breast cancer. Journal of Personality and Social Psychology, 65 375-390. Carver, C. S., Scheier, M. F., & Weintraub, J. K. (1989). Assessing coping strategies: A theoretically based approach. Journal of Personality and Social Psychology, 56 (2), 267-283. Chang, E. C., Maydeu-Olivares, A., & DÂ’Zurilla, T. J. (1997). Optimism and pessimism as partially independent constructs: Relations to positive and negative affectivity and psychological well-being. Personality and Indivi dual Differences, 23 433-440. Chang, E. C., & Sanna, L. J. (2001). Optimism, pessimism, and positive and negative affectivity in middle-aged adults: A test of a cogniti ve-affective model of psychological adjustment. Psychology and Aging, 16 (3), 524-531. Chang, E. C., Sanna, L. J., & Yang, K-M. (2003). Optimism, pessimism, affectivity, and psychological adjustment in US and Korea: A test of a mediation model. Personality and Individual Differences, 34 1195-1208. Cheng, S., & Chan, A. C. M. (2005). The Center for Epidemiologic Studies Depression Scale in older Chinese: Th resholds for long and short forms. International Journal of Geriatric Psychiatry, 20 (5), 465-470. Chatters, L. M. (2000). Religion and health : Public health res earch and practice. Annual Review of Public Health, 21 335-367.
205 Cochran, J. K., Beeghley, L., & Bock, E. W. (1988). Religiosity and alcohol behavior: An exploration of reference group theory. Sociological Forum, 3 256-276. Cohen, M. J. (1989). Employment and volunteering roles for the elderly: Characteristics, attributions, and strategies. Journal of Leisure Research, 21 (3), 214-227. Cohen, S., & Willis, T. A. (1985). Stress, social support, and the buffering hypothesis. Psychological Bulletin, 98 310-357. Coleman, D., & Iso-Ahola, S. (1993). Leisure and health: The role of social support and self-determination. Journal of Leisure Research, 25 111-128. Commerford, M. C., & Reznikoff, M. (1996). Relationship of religion and perceived social support to self-esteem a nd depression in nursing home residents. The Journal of Psychology, 130 35-50. Coughlin, J. F. (1999). Technology needs of aging Baby Boomers. Issues in Science and Technology Online, Fall [Online]. Covinsky, K. E., Kahana, E., Chin, M. H., Palmer, R. M., Fortinsky, R. H., & Landefeld, C. S. (1999). Depressive sy mptoms and 3-year mortality in older hospitalized medical patients. Annals of Internal Medicine, 130 (7), 563-569. Cowen, E. L. & Work, W. C. (1988). Resilient children, psychological wellness, and primary prevention. American Journal of Community Psychology, 16( 4), 591-607. Cox, H., & Hammonds, A., (1988). Religio sity, aging and life satisfaction. Journal of Religion and Aging 5 (2), 1-21. Crandall, J. E. (1975). A scale for social interest. Journal of Individual Psychology, 31 187-195.
206 Crimmins, E. M., Saito, Y., Reynolds, S. L. (1997). Further evidence on recent trends in the prevalence and incidence of disability among older Americans from two sources: The LSOA and the NHIS. Journal of Gerontology: Social Science, 52B 59-71. Cronbach, L. J. (1951). Coefficient alpha and the internal stru cture of tests. Psychometrika, 16 297-334. Crowther, M. R., Parker, M. W., Ache nbaum, W.A., Larimore, W. L., & Koenig, H. G. (2002). Rowe and Kahn's model of successf ul aging revisited: Positive spirituality the forgotten factor. The Gerontologist, 42 (5), 613-620. Cumming, E., & Henry, W.E. (1961). Growing old: The process of disengagement New York: Basic Books. Cutler, S. J. (1976). Membership in diffe rent types of voluntary associations and psychological well-being. The Gerontologist, 16 335-339. Cutler, S., & Danigelis, N. (1993). Orga nizational contexts of activity. In: J. Kelly (Ed.), Activity and Aging pp. 146-163. Newbury Park, CA: Sage. Cutrona, C. E. (1986). Objective dete rminants of perceived social support. Journal of Personality and Social Psychology, 50 349-355. Davis, B. (2005). Mediators of the re lationship between hope and well-being in older adults. Clinical Nursing Research, 14 (3), 253-272. Diener, E. (1984). Subjective well-being. Psychological Bulletin, 93 542-575. Diener, E., & Emmons, R. A. (1984). The independence of positive and negative affect. Journal of Personality and Social Psychology, 47 1105-1117. Diener, E., Emmons, R. A., Larsen, R. J ., & Griffin, S. (1985). The Satisfaction With Life Scale. Journal of Personality Assessment, 49 71-75.
207 Diener, E., Sandvik, E., & Larsen, J. (1985). Age and sex effects for affect intensity. Developmental Psychology, 21 542-546. Diener, E., Sandvik, E., Pavot, W., & Fujita, F. (1992). Extraversion and subjective well-being in a U.S. national probability sample. Journal of Research in Personality, 26 205-215. Diener, E., Suh, E. M., Lucas, R. E., & Smith, H. L. (1999). Subjective wellbeing: Three decades of progress. Psychological Bulletin, 125 276-302. Donahue, M. J. (1985). Intrinsic and ex trinsic religiousness: Review and metaanalysis. Journal of Personality and Social Psychology, 48 400-419. Dorfman, L. T., Kohut, F. J., & Heckert, D. A. (1985). Retirement satisfaction in the rural elderly. Research on Aging, 7 577-599. Duke University Center for the Stud y of Aging and Human Development. (1978). Multidimensional Functional Asse ssment: The OARS Methodology Durham, NC: Duke University. Dull, V. T., & Skokan, L. A. (1995). A cognitive model of religionÂ’s influence on health. Journal of Social Issues, 51 (2), 49-65. Durkheim, E. (1933). The division of labor in society. New York: Free Press. (Original work published 1901). Dyer, J. G., & McGuiness, T.M. (1996). Resilience: Analysis of the concept. Archives of Psychiatric Nursing, 10 276-282. Edwards, J. M., & Klemmack, D. L. (1973) Correlates of life satisfaction: A reexamination. Journal of Gerontology, 28 497-502.
208 Egeland, B., Carlson, E., & Sroufe, L. A. (1993). Resilience as a process. Development and Psychopathology, 5 517-528. Eisenberg, D. M., Kessler, R. C., Foster, C., Norlock, F. E., Calkins, D. R., & Delbanco, T. L. (1993). Unconventional me dicine in the United States: Prevalence, costs, and patterns of use. New England Journal of Medicine, 328 (4), 246-252. Ellison, C. (1991). Religious involvement and subjective well-being. Journal of Health and Social Behavior, 32 80-99. Ellison, C. (1994). Religion, the life stress paradigm, and the study of depression. In : J. S. Levin: Religion in Aging and Health, Theoretical Foundations and Methodological Frontiers (pp. 78-121). Thousand Oaks, CA : Sage Publications, Inc. Ellison, C. G. (1995). Race, religious involvement and depressive symptomatology in a southeastern U. S. community. Social Science & Medicine, 40 1561-1572. Ellison, C. G., & Levin, J. S. (1998). The religion-health connection: Evidence, theory, and future directions. Health Education and Behavior, 25 700-720. Elliot, T. R., Marmarosh, C., & Pickelman, H. (1994). Negative affectivity, social support, and the prediction of depression and distress. Journal of Personality, 62 299319. Eng, E., Hatch, J., & Callan, A. (198 5). Instutionalizing social support through the church and into the community. Health Education Quarterly, 12 81-92. Erikson, E.H. (1959). Identity and the life cycle. Psychological Issues, 1 1-171 Feinson, M. C. (1985). Aging and me ntal health: Distingushing myth from reality. Research on Aging, 7 155-174.
209 Feldman, J. J., Makuc, D. M., Kleinman J. C., & Cornoni-Huntley, J. (1989). National trends in educational differentials in mortality. American Journal of Epidemiology, 129 919-939. Felten, B. S. (2000). Resilience in a multicultural sample of community-dwelling women older than age 85. Clinical Nursing Research, 9 (2), 102-123. Ferraro, K. F., & Koch, J. R. (1994) Religion and health among Black and White adults: Examining social support and consolation. Journal for the Scientific Study of Religion, 33 362-375. Field, D., & Minkler, M. (1988). C ontinuity and change in social support between young-old, old-old, and very-old adults. Journal of Gerontology: Psychological Sciences, 43 P100-P106. Fillenbaum, G. G. (1978). Multidimensional Functional Assessment: The OARS Methodology (Second Addition). Durham, N. C.: Duke University. Fillenbaum, G. G. (1979). Social context and self-assessments of health among the elderly. Journal of Health and Social Behavior, 20 45-51. Fleury, J. (1998). The index of self-re gulation: Development and psychometric analysis. Journal of Nursing Measurement, 6 (1), 3-17. Ford, A. B., Haug, M. R., Jones, P. K., Roy, A. W., & Folmar, S. J. (1990). Race-related differences among elderly urban residents: A cohort study. Journal of Gerontology, Social Sciences, 45 S163-S171. Ford, A. B., Haug, M. R., Stange, K. C., Gaines, A. D., Noelker, L. S., & Jones, P. K. (2000). Sustained personal autono my: A measure of successful aging. Journal of Aging and Health, 12 (4), 470-489.
210 Foster, J. R. (1997). Successful coping, adaptation and resilience in the elderly: An interpretation of epidemiologic data. Psychiatric Quarterly, 68 189-219. Fratiglioni, L., Wang, H. X., Ericsson, K., Maytan, M., & Winblad, B. (2000). Influence of social networks on occurrence of dementia: A community-based longitudinal study. The Lancet, 355 1315-1319. Fredrickson, B. L. (2001). The role of positive emotions in positive psychology. American Psychologist, 56 (3), 218-226. Fredrickson, B. L., & Joiner, T. (2002) Positive emotions trigger upward spirals toward emotional well-being. Psychological Sciences, 13 172-175. Freund, A..M. & Baltes, P.B. (1998). Selection, optimization, and compensation as strategies of life management: Correlati on with subjective indicators of successful aging. Psychology and Aging, 13 (4), 531-543. Friedman, L. C., Nelson, D. V., Baer, P. E., Lane, M., Smith, F. E. & Dworkin, R. J. (1992). The relationship of dispositiona l optimism, daily life stress, and domestic environment to coping methods used by cancer patients. Journal of Behavioral Medicine, 15 127-142. Fries, J. E. (1980). Aging, natural death, and the compre ssion of morbidity. The New England Journal of Medicine, 303 (3), 130-135. Fries, J. F. (2003). Measuring and monitoring success in compressing morbidity. Annals of Internal Medicine, 139 455-459.
211 Fung, H. H., Carstensen, L. L., & Lang, F. R. (2001). Age-related patterns in social networks among European Americans a nd African Americans: Implications for socioemotional selectivity across the life span. International Journal of Aging and Human Development, 52 (3), 185-206. Futterman, A. D., Kemeny, M. E., Shapiro, D., & Fahey, J. L. (1994). Immunological and physiological changes associ ated with induced positive and negative mood. Psychosomatic Medicine, 56 499-511. Gallup, G. (1995). The Gallup poll: Public Opinion, 1995 Wilmington, DE: Scholarly Resources. Garmezy, N. (1985). Stress resistant childr en: The search for protective factors. Journal of Child Psychology and Psychiatry, 4 213-233. Garmezy, N. (1991). Resilience in child renÂ’s adaptation to negative life events and stressed environments. Pediatric Annals, 20 459-466. Garmezy, N. (1993). Resilience despite risk. Psychiatry, 56 ,127-136. George, L. K. (1981). Subjective wellbeing: Conceptual and methodological issues. Annual Review of Gerontology and Geriatrics, 2 345-382. George, L. K. (1996). Social factors a nd illness. In R. H. Binstock & L. K. George (Eds). Handbook of Aging and the Social Sciences Fourth Edition. San Diego, CA: Academic Press. George, L. K., & Clipp, E. (1991). Subjective components of aging well. Generations, 15 57-60.
212 George, L. K. & Fillenbaum, G. G. (1985). OARS Methodology: A decade of experience in geriatric assessment. Journal of the American Geriatrics Society, 33 607615. Glass, T. A., Seeman, T. E., Herzog, R. A., Kahn, R., & Berkman, L. F. (1995). Change in productive activity in late adulthood : MacArthur Studies of Successful Aging. Journal of Gerontology, Medical Sciences, 50B 565-576. Gold, M., Franks, P., & Erickson, P. ( 1996). Assessing the hea lth of a nation: the predictive validity of a preferencebased measure and self-rated health. Medical Care, 34 (2), 163-177. Gruenberg, E. M. (1977). The failure of success. Milbank Memorial Fund Quarterly, 55 3-24. Gubrium, J. (1972). Toward a socio-environmental theory of aging. The Gerontologist, 12 281-284. Guralnik, J. M., Fried, L. P., & Salive, M. E. (1996). Disability as a public health outcome in the aging population. Annual Review of Public Health, 17 25-46. Guralnik, J. M., LaCroix, A. Z., Abbo tt, R. D., et al. (1993). Maintaining mobility in late life. I Demographich characteristics and chronic conditions. American Journal of Epidemiology, 137 845-857. Guttman, C. (2000). Older Americans 2000: New data system tracks health and well-being finds successes and disparities. Geriatrics, 55 (10), 63-69. Haan, M., Kaplan, G. A., & Camacho, T. (1987). Poverty and health: Prospective evidence from the Alameda County Study. American Journal of Epidemiology, 125 989-998.
213 Haber, D. J. (1984). Church-based programs for Black care givers of noninstitutionalized elders. Journal of Gerontological Social Work, 7 (4), 43-56. Hackney, C. H., & Sanders, G. S. (2 003). Religiousness and mental health: A meta-analysis of recent studies. Journal for the Scientifi c Study of Religion, 42 43-55. Hadaway, C. K., & Roof, W. C. (1978) Religious commitment and the quality of life in American society. Review of Religious Research, 19 295-307. Hardy, S. E., Concato, J., & Gill, T. M. (2004). Resilience of communitydwelling older persons. Journal of the American Geriatrics Society, 52 257-262. Hatcher, L., & Stepanski, E. J. (1994). A step-by-step approach to using the SAS System for univariate and multivariate statistics Cary, NC: SAS Institute, Inc. Havighurst, R.J. (1961). Successful aging. The Gerontologist, 1 8-13. Hays, J. C., Landerman, L. R., Blazer, D. G., Koenig, H. G., Carroll, J. W., & Musick, M. A. (1998). Aging, heal th, and the Â“electronic church.Â” Journal of Aging and Health, 10 458-482. Herzog, A. R., & House, J.S. (1991) Productive activitie s and aging well. Generations, 15 49-54. Hill, P. C., & Pargament, K. I. (2003) Advances in the conceptualization and measurement of religion and spirituality. American Psychologist, 58 (1), 64-74. Hillers, P. K., Jorm, A. F., Herlitz, A., & Winblad, B. (1998). Negative and positive affect among the very old. Research on Aging, 20 (5), 593-610. Hirsch, M. (1999). A role for adapte d physical educators in gerontology. Palestra, 15 (2), 5.
214 Hobfoll, S. E. (1989). Conservativa tion of resources: A new attempt at conceptualizing stress. American Psychologist, 44 513-524. Hochschild, A. R. (1975). Disengageme nt theory: A critique and proposal. American Sociological Review, 40 553-569. Hodges, S. (2002). Mental health, depre ssion, and dimensions of spirituality and religion. Journal of Adult Development, 9 (2), 109-115. Hogan, D., Eggebeen, D., & Clogg, C. ( 1993). The structure of intergenerational exchanges in American families. American Journal of Sociology, 98 1428-1458. Hoge, D. R. (1996). Religion in Amer ica: The demographics of belief and affiliation. In E. P. Shafranske (Ed.), Religion and the clinical practice of psychology (pp. 21-41). Washington, DC: Amer ican Psychological Association. Holahan, C. K., & Holahan, C. J. (1987). Self-efficacy, social support, and depression in aging: A longitudinal analysis. Journal of Gerontology, 42 65-68. Hooyman, N., & Kiyak, H. A. (1999). Social Gerontology. A Multidisciplinary Perspective (5th ed.). Boston, MA: Allyn & Bacon. House, J. S., Kessler, R. C., & Herzog, A. R. (1990). Age, socioeconomic status, and health. The Milbank Quarterly, 68 383-411. House, J. S., Robbins, C., & Metzner, H. L. (1982). The association of social relationships and activities with mortality: Prospective evidence from the Tecumseh Community Health Study. American Journal of Epidemiology, 116 123-140. Hummer, R., Rogers, R. G ., Nam, C. B., & Ellison, C. G. (1999). Religious involvement and U. S. adult mortality. Demography, 36 273-285.
215 Hunsberger, B. (1985). Religion, age, life satisfaction, and perceived sources of religiousness: A study of older persons. Journal of Gerontology, 40 615-620. Idler, E. (1987). Religious involvement and the health of the elderly: Some hypotheses and an initial test. Social Forces, 66 226-238. Idler, E. L., & Kasl, S. V. (1991). H ealth perceptions and survival: Do global evaluations of health status really predict mortality? Journal of Gerontology: Social Sciences, 46 (2), S55-S65. Idler, E. L., & Kasl, S. V. (1992). Relig ion, disability, depression, and the timing of death. American Journal of Sociology, 97 (4). 1052-1079. Idler, E. L., & Kasl, S. V. (1997). Religion among disabled and nondisabled persons I: Cross-sectional patterns in health practices, social activities, and well-being. Journal of Gerontology: Social Sciences, 52B S294-S305. Idler, E. L., Kasl, S. V., & Lemke, J. H. (1990). Self-evaluated health and mortality among the elderly in New Haven, Connecticut, and Iowa and Washington Counties, Iowa, 1982-1986. American Journal of Epidemiology, 17 (1), 91-103. Idler, E. L., Musick, M.A., Ellison, C.G ., George, L.K., Krause, N., et. al. (2003). Measuring multiple dimensions of religion and spirituality for health research; Conceptual background and findings fr om the 1998 General Social Survey. Research on Aging, 25( 4), 327-365. Inui, T. S. (2003). The need for an integrated biopsychosocial approach to research on successful aging. Annals of Internal Medicine, 139 391-394.
216 Irwin, M., Artin, K. H., & Oxman, M. N. (1999). Screening for depression in the older adult: criterion validity of the 10-item Center for Epidemiological Studies Depression Scale (CES-D). Archives of Internal Medicine, 159( 15), 1657-1658. Jacelon, C. S. (1997). The tra it and process of resilience. Journal of Advanced Nursing, 25 (1), 123-129. James, L. R., & Brett, J. M. (1984). Mediators, moderators, and tests for mediation. Journal of Applied Psychology, 69, 307-321. Jolly, J. B., Dyck, M. J., Krammer, T. A ., & Wherry, J. N. (1994). Integration of positive and negative affectivity and c ognitive content-specificity: Improved discrimination of anxious and depressive symptoms. Journal of Abnormal Psychology, 103 544-552. Jylh, M., & Aro, S. (1989). Social ties and survival among the elderly in Tampere, Finland. International Journal of Epidemiology, 18, 158-164. Kahana, E. & Kahana, B. (2002). Cont extualizing successful aging: New directions in an age-old search In J. R. Settersten (Ed.), Invitation to the Life Course: A New Look at Old Age (pp. 225-255). Amityville, NY: Ba ywood Publishing Company. Kahana, E., Kahana, B., & Kercher, K. (1999). Emerging lifestyles and proactive options for successful aging. Ageing International, 28 (2), 155-189. Kahana, E., Lawrence, R.H., Kahana, B., Ke rcher, K., Wisniewski, A., Stoller, E., Tobin, J., & Stange, K. (2002). Long-term impa ct of preventive proactivity on quality of life of the old-old. Psychosomatic Medicine, 64 382-394.
217 Kahn, J. H., Hessling, R. M., & Russell, D. W. (2003). Social support, health, and well-being among the elderly: What is the role of negative affectivity? Personality and Individual Differences, 35, 5-17. Kahn, R. L., & Antonucci, T. C. (1 980). Convoys over the life course: Attachment, roles and social support. In P. B. Baltes & O. Brim (Eds), Aging: Biology, and behavior (Vol. 3, pp. 254-283). Boston: Lexington. Kaplan, G. A. & Camacho, T. (1983). Pe rceived health and mo rtality: A nineyear follow-up of the human population laboratory cohort. American Journal of Epidemiology, 117 (3), 292-304. Kark, J.D., Shemi, G., Friedlander, Y ., Martin, O., Manor, O., & Blondheim, S. H. (1996). Does religious observance promote health? Mortality in Secular vs. religious kibbutzim in Israel. American Journal of Public Health, 86 341-346. Kercher, K. (1992). Assessing subjective we ll-being in the old-old. The PANAS as a measure of orthogonal dimensions of positive and negative affect. Research on Aging, 14 (2), 131-168. Kerschner, H., & Pegues, J. (1998). Produc tive aging: A qualit y of life agenda. Journal of the American Dietetic Association, 8 (2), 445. Kivela, S. L., & Pahkala, K. (2001). De pressive disorder as a predictor of physical decline in old age. Journal of the American Geriatric Society, 49 290-296. Knapp, M. R. J. (1977). The Activity Theory of aging. The Gerontologist, 17 (6), 553-559. Kobasa, S. C. (1979). Stressful life ev ents, personality and health: An inquiry into hardiness. Journal of Personality and Social Psychology, 37 707-717.
218 Koenig, H. G. (1990). Research on religion and mental health in later life: A review and commentary. Journal of Geriatric Psychiatry, 23 23-53. Koenig, H. G. (1995). Use of acute hospital services and mortality among religious and non-religious copers with medical illness. Journal of Religious Gerontology, 9 (3), 1-21. Koenig, H. G. (2001). Religion and medici ne II: Religion, mental health and related behaviors. International Journal of Psychiatry in Medicine, 31 97-109. Koenig, H. G. (2004). Featured CME topic: Spirituality. Southern Medical Journal, 97 (12), 1194-1200. Koenig, H. G., George, L. K., & Siegler, I. C. (1988). The use of religion and other emotion-regulating coping st rategies among older adults. The Gerontologist, 28 18-28. Koenig, H. G., Kvale, J. N., & Ferrel, C. (1988). Religion and well-being in later life. The Gerontologist, 28 18-28. Koenig, H. G., & Larson, D. B. (1998). Use of hospital services, religious attendance, and religious affiliation. Southern Medical Journal, 91 (10), 925-932. Kohout, F. J., Berkman, L. F., Evans, D. A., & Cornoni-Huntley, J. (1993). Two shorter forms of the CES-D Depression Index. Journal of Aging and Health, 5 (2), 179193. Kosmin, B. A., & Lachman, S. P. (1993). One nation under God New York: Harmony. Krach, P., DeVaney, S., DeTurk, C., & Zi nk, M. (1996). Functional status of the oldest-old in a home setting. Journal of Advanced Nursing, 24 (3), 456-464.
219 Krause, N. (1986). Stress and coping: Reconceptualizing the role of locus of control beliefs. Journal of Gerontology, 41 617-622. Krause, N. (1990). Perceived health pr oblems, formal/informal support, and life satisfaction among older adults. Journal of Gerontology: Social Sciences, 5 S193-205. Krause, N. (1992). Stress, religiosity, and psychological well-being among older Blacks. Krause, N. (1993). Measuring re ligiosity in later life. Research on Aging, 15 (2), 170-197. Krause, N. (1995). Negative interacti on and satisfaction with social support among older adults. Journal of Gerontology: Psychological Sciences, 50B P59-P73. Krause, N. (1998). Neighborhood deteriora tion, religious coping, and changes in health during late life. The Gerontologist, 38 653-664. Krause, N., & Tran, T. V. (1989). Stre ss and religious involvement among older Blacks. Journal of Gerontology: Social Sciences, 44, 4-13. Kutner, N. F., Ory, M. G., Baker, D. I., Schechtman, K. B., Hornbrook, M. C., & Mulrow, C. D. (1992). Measuring the quality of life of the elderly in health promotion intervention clinical trials. Public Health Reports, 107 530-539. Lachman, M. E., & Weaver, S. L. (1998). The sense of control as a moderator of social class differences in health and well-being. Journal of Personality and Social Psychology, 74 763-773. LaFerriere, R. H., & Hamel-Bissell, B. P. (1994). Successful aging of the oldest old women in the northeast kingdom of Vermont. IMAGE: Journal of Nursing Scholarship, 26 319-323.
220 Larson, R. (1978). Thirty years of res earch on the subjective well-being of older Americans. Journal of Gerontology, 33 (1), 109-125. Larson, E. B. (1997). An overview (Successful aging). The Western Journal of Medicine, 167 (4), 204-206. LaRue, Bank, Jarvik, & Hetland. (1979). Health in old age: How do physiciansÂ’ ratings and self-ratings compare? Journal of Gerontology, 34 687-691. Laukkanen, P., Leskinen, E., Kauppinen, M., Sakari-Rantala, R., & Heikkinen, E. (2000). Health and functional capacity as predictors of community dwelling among elderly people. Journal of Clinical Epidemiology, 53 257-265. Lawton, M. P. (1985). Activities and leisure. Annual Review of Gerontology and Geriatrics, 5 127-164. Lawton, M. P., & Brody, E. M. (1969). Assessment of older people: Selfmaintaining and instrumental activities of daily living. The Gerontologist, 9 179-186. Lee, R. D., & Carter, L. R. (1992). Modeling and forecasting U.S. mortality. Journal of the American St atistical Association, 87 (419), 659-671. Lee, D. J., & Markides, K. S. (1990). Activity and mortality among aged persons over an eight-year period. The Journals of Gerontology: Social Sciences, 45 S39-S42. Lemon, B.W., Bengston, V.L., & Peterson, J.A. (1972). An exploration of the Activity Theory of aging: Activity types and life satisfaction among in-movers to a retirement community. Journal of Gerontology, 27 511-523. Lennartsson, C., & Silverstein, M. (2001) Does engagement with life enhance survival of elderly people in Sweden? Th e role of social and leisure activities. Journal of Gerontology: Social Sciences, 56B (6), S335-S342.
221 Leveille, S. G., Guralnik, J. M., Ferru cci, L., & Langlois, J. A. (1999). Aging successfully until death in old age: Opportuni ties for increasing active life expectancy. American Journal of Epidemiology, 149 (7), 654-664. Levin, J. S. (1994a). Investigating the epidemiologic effects of religious experience: Findings, explanations, a nd barriers. In: J. S. Levin: Religion in Aging and Health, Theoretical Foundations and Methodologi cal Frontiers (pp. 3-17). Thousand Oaks, CA: Sage Publications, Inc. Levin, J. S. (1994b). Religion and health: Is there an association, is it valid, and is it causal? Social Science and Medicine, 38 1475-1482. Levin, J. S., & Markides, K. S. ( 1985). Religion and health in Mexican Americans. Journal of Religion and Health, 24 ( 1), 60-69. Levin, J. S., & Schiller, P. L. (1987). Is there a religious factor in health? Journal of Religion and Health, 26 9-36. Levin, J. S., & Taylor, R. J. (1998). Pane l analyses of religious involvement and well-being in African Americans: Contemporaneous vs. longitudinal. Journal for the Scientific Study of Religion 37(4), 695-709. Levin, J. S., Taylor, R. J., & Chatters, L. M. (1994). Race and gender differences in religiosity among older adults: Find ings from four national surveys. Journal of Gerontology, 49 S137-S145. Levin, J. S., & Vanderpool, H. Y. (1987). Is frequent religio us attendance really conducive to better health? Toward an epidemiology of religion. Social Science and Medicine, 24 589-600.
222 Levin, J. S., & Vanderpool, H. Y. (1989). Is religion therapeutically significant for hypertension? Social Science and Medicine, 29 69-78. Lewinsohn, P. M., Seeley, J. R., Roberts, R. E., & Allen, N. B. (1997). Center for Epidemiologic Studies Depression Scale (C ES-D) as a screening instrument for depression among community-residing older adults. Psychology and Aging, 12 (2), 277287. Lin, T. H. (1997). A SAS program for testing the statistical significance of a mediating variable. Education and Psycholgical Measurement, 57 (1), 186-188. Locke, L. F., Wyric Spirduso, W., & Silverman, S. J. (2000). Proposals that Work: A Guide for Planning Dissertations and Grant Proposals (4 ed.). Thousand Oaks: Sage Publications, Inc. Longino, C. F., & Kart, C. S. (1982). Explicating Activity Theory: A formal replication. Journal of Gerontology, 37 (6), 713-722. Lucas, R. E., Diener, E. & Suh, E. (1996) Discriminant validity of well-being measures. Journal of Personality and Social Psychology, 71, 616-628. Lubitz, J., Beebe, J., & Baker, C. (1995). Longevity and Medicare expenditures. The New England Journal of Medicine, 332 (15), 999-1003. Lucas, R. E., Diener, E., & Suh, E. (19 96). Discriminant validity of well-being measures. Journal of Personality and Social Psychology, 71 (3), 616-628. Luthar, S. S. & Cicchetti, D. (2000). The c onstruct of resilience: Implications for interventions and social policies. Development and Psychopathology, 12 857-885. MacKinnon, D. P., & Dwyer. (1993). Estim ating mediated effects in prevention studies. Evaluation Review, 17 144-158.
223 MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V. (2002). A comparison of methods to test media tion and other interveni ng variable effects. Psychological Methods, 7 83-104. MacKinnon, D. P., Warsi, G., & Dwyer, J. H. (1995). A simulation study of mediated effects measures. Multivariate Behavioral Research, 30 (1), 41-62. MacKinnon, D. P., Warsi, G., & Dwyer, J. H. (1995). A simulation study of mediated effects measures: Erratum. Multivariate Behavioral Research, 30 (3), ii. Maddox, G. L. (1962). Some correlates of differences in self-assessment of health status among the elderly. Journal of Gerontology, 17 180-185. Maddox, G. L. (1987). Aging differently. The Gerontologist, 27 557-564. Maddox, G. L., & Douglass, E. B. (1973). Self-assessment of health: A longitudinal study of elderly subjects. Journal of Health and Social Behavior, 14 87-93. Major, B., Richards, C., Cooper, M. L., Cozzarelli, C., & Zubek, J. (1998). Personal resilience, cognitive appraisals, and co ping: An integrative model of adjustment to abortion. Journal of Personality and Social Psychology, 74 (3), 735-752. Mangen, D. J. & Peterson, W. H. (1982). Research Instruments in Social Gerontology, Vol 1: Clinical and Social Psychology. Minne apolis, MN: University of Minnesota Press. Marmot, M. g., Kogevinas, M., & elston, M. A. (1987). Social/economic status and disease. Annual Review of Public Health, 8 111-135.
224 Marshall, G. N., Wortman, C. B., Kusulas, J. W., Hervig, L. K., & Vickers, Jr. R. R. (1992). Distinguishing optimism from pessimism: Relations to fundamental dimensions of mood and personality. Journal of Personality and Social Psychology, 62 1067-1074. Masten, A.S., & OÂ’Connor, M.J. (1989). Vulnerability, stress, and resilience in the early development of a high risk child. Journal of the American Academy of Child Adolescent Psychiatry, 28 274-278. Mattlin, J. A., Wethington, E., & Kessler, R. C. (1990). Situational determinants of coping and coping effectiveness. Journal of Health and Social Behavior, 31 103-122. Mattis, J. S., Fontenot, D. L., & Hatcher-K ay, C. A. (2003). Religiosity, racism, and dispositional optimism among African Americans. Personality and Individual Differences 34, 1025-1038. McCrae, R. R. (1984). Situational dete rminants of coping responses: Loss, threat and challenge. Journal of Personality and Social Psychology, 46 919-928. McCrae, R. R. & Costa, P. T. (1988). Psychological resilience among widowed men and women: 10-year follow-up of a national sample. Journal of Social Issues, 44 129-142. McCullough, M. E., Larson, D. B., Hoyt, W. T., Koenig, H. G., & Thoresen, C. (2000). Religious involvement and mort ality: A meta-analytic review. Health Psychology, 19 (3), 211-222. McIntosh, W. A., & Alston, J. P. (1982). Lenski revisited: The linkage role of religion in primary and secondary groups. American Journal of Sociology, 87 852-882.
225 McHorney, C. A., Ware, J. E. Jr., & R aczek, A. E. (1993). The MOS 36-Item Short-Form Health Survey (SF-36) II: Psyc hometric and clinical tests of validity in measuring physical and mental health constructs. Medical Care, 31 247-263. Medicare Trustees. (2003). 2003 annual report of the Bo ards of Trustees of the Federal Hospital Insurance Trust and Federal Supplementary Medical Insurance Trust Funds. Washington, DC: Centers for Medicare and Medicaid Services. Miller, T. (2001). Increasing longe vity and Medicare expenditures. Demography, 38 (2), 215-226. Mills, T. L. (2001). Comorbid depressive symptomatology: Isolating the effects of chronic medical conditions on self-repor ted depressive symptoms among communitydwelling older adults. Social Science & Medicine, 53 569-578. Minkler, M. & Fadem, P. (2002). Successful aging: A disability perspective. Journal of Disability Policy Studies, 12 (4), 229-236. Moen, P., Dempster-McClain, D., & Williams, R. M. Jr. (1989). Social integration and longevity: An event history analysis of wo menÂ’s roles and resilience. American Sociological Review 54, 635-647. Moen, P. & Erickson, M.A. (1995). Linke d lives: A transgen erational approach to resilience. In : Phyllis Moen, Glen H. Elder, Jr., & Kurt Lscher (Eds.), Examining lives in context: Perspectives on the ecology of human development (pp.169-210). Washington, DC: American Psychological Association. Morgan, G. A., Vaske, J. J., Gliner, J. A., & Harmon, R. J. (2003). Logistic regression and discriminant analys is: Use and interpretation. Journal of the American Academy of Child & Adolescent Psychiatry, 42 (8), 994-997.
226 Morris, D. C. (1991). Church attendance, religious activities and life satisfaction of older adults in Middletown, U.S.A. Journal of Religious Gerontology, 81 (1), 83-96. Morse, C. K., & Wisocki, P. A. (1988) Importance of religiosity to elderly adjustment. Journal of Religion and Aging, 4 15-26. Mossey, J. M. & Shapiro, E. (1982). Se lf-rated health: A pr edictor of mortality among the elderly. American Journal of Public Health, 72 (8), 800-808. Mulholland, D. J., Watt, N. F., Philpott, A., & Sarlin, N. (1991). Academic performance in children of divorce: Ps ychological resilience and vulnerability. Psychiatry, 54 268-280. Munro, B. H. (2001). Statistical methods for health care research, 4th ed Philadelphia, PA: Lippincott Williams & Wilkins. Musick, M. A. (1996). Religion and subjective health among Black and White elders. Journal of Health and Social Behavior, 37 221-237. Musick, M. A., Traphagan, J. W., Koenig, H. G., & Larson, D. B. (2000). Spirituality in physical health and aging. Journal of Adult Development, 7 (2), 73-86. Myers, G. C., & Manton, K. G. (1984) Compression of mortality: Myth or reality? Gerontologist, 24 346-53. Nakashima, M., & Canda, E. R. (2005). Posi tive dying and resilien cy in later life: A qualitative study. Journal of Aging Studies, 19 109-125. Neill, C. M., & Kahn, A. S. (1999). The role of personal spirituality and religious social activity on the life satisfaction of older widowed women. Sex Roles, 40 319-329.
227 Neugarten, B. L., Havighurst, R. J., & T obin, S. (1961). The measurement of life satisfaction. Journal of Gerontology, 16 134-143. Newsom, J. T., & Schulz, R. (1996). Soci al support as a mediator in the relation between functional status and qual ity of life in older adults. Psychology and Aging, 11 (1), 34-44. Nunnally, J. (1978). Psychometric theory New York: McGraw-Hill. Okun, M. A., & Keith, V. M. (1998). Effects of positive and negative social exchanges with various sources of depressi ve symptoms in younger and older adults. Journal of Gerontology: Psychological Sciences, 53B P4-P20. Okun, M. A., & Stock, W. A. (1987) Correlates and components of subjective well-being among the elderly. Journal of Applied Gerontology, 6, 95-112. Okun, M. A., Stock, W. A., Haring, M. J., & Witter, R. A. (1984). The social activity/subjective well-being relation: A quantitative synthesis. Research on Aging, 6 45-65. Olsen, R. B., Olsen, J., Gunner-Svensson, F., & Waldstrom, B. (1991). Social networks and longevity: A 14 year followup study among the elderly in Denmark. Social Science & Medicine, 33 1189-1195. Oman, D., & Reed, D. (1998). Religion and mortality among the communitydwelling elderly. American Journal of Public Health, 88 (10), 1469-1475. Ortega, S. T., Crutchfield, R. D., & Rush ing, W. A. (1983). Race differences in elderly personal well-being: Fr iendship, family and church. Research on Aging, 4 101117.
228 Orth-Gomer, K., & Johnson, J. V. (1987). Social network interaction and mortality. A six year follow-up study of a random sample of the Swedish population. Journal of Chronic Disease, 40 949-957. Orth-Gomr, K., Rosengren, A., Wilhelmsen L. (1993). Lack of social support and incidence of coronary heart disease in middle-aged Swedish men. Psychosomatic Medicine 55, 37-43. Ostir, G. V., Markides, K. S., Black, S. A., & Goodwin, J. S. (2000). Emotional well-being predicts subsequent func tional independence and survival. Journal of the American Geriatrics Society, 48 473-478. Oxman, T. E., Freeman, D. H., & Manheimer, E. D. (1995). Lack of social participation or religious strength and comf ort as risk factors for death after cardiac surgery in the elderly. Psychosomatic Medicine, 57 5-15. Palmore, E.B. (1979). Predictors of successful aging. The Gerontologist, 19 (5), 427-31. Pargament, K. I., Ensing, D. S., Falgout, K., Olsen, H., Reilly, B., et al. (1990). God help me: Religious coping efforts as predictors of the outcomes to significant negative life events. American Journal of Community Psychology, 18 793-824. Parmelee, P. A., Katz, I. R., & Lawton, M. P. (1989). Depression among institutionalized aged: Assessment and preval ence estimation. Journals of Gerontology, 33, 22-24. Parsons, T. (1951). The Social System Glencoe, IL: The Free Press.
229 Patrick, D. L., Kinne, S., Engelberg, R. A., & Pearlman, R. A. (2000). Functional status and perceived quality of life in a dults with and without chronic conditions. Journal of Clinical Epidemiology, 53 779-785. Pavot, W., & Diener, E. (1993). Review of the satisfaction with life scale. Psychological Assessments, 5 (2), 164-172. Pavot, W., Diener, E., Colvin, C. R., & Sa ndvik, E. (1991). Further validation of the Satisfaction With Life Scale: Evidence for the cross-method convergence of wellbeing measures. Journal of Personality Assessment, 57 149-161. Pearlin, L. (2001). Bereavement and the loss of mattering. In T. J. Owens & S. Stryker (Eds.), Extending self-esteem theory and research: Sociological and psychological currents pp. 285-300. New York, NY: Cambridge University Press. Pearlin, L. I., & Schooler, C. (1978). The structure of coping. Journal of Health and Social Behavior, 19 2-21. Pedhazur, E. J., & Pedhazur Schmelkin, L. (1991). Measurement, design, and analysis: An integrated approach. Hilldale, N J: Lawrence Erlbaum Associates. Penninix, B. W. J. H., Guraln ik, J. M., Ferrucci, L., Simonsick, E. M., Deeg, D. J. H., & Wallace, R. B. Depressive symptoms and physical decline in community dwelling older persons. Journal of the American Medical Association, 279 1720-1726. Peterson, C. (2000). The future of optimism. American Psychologist, 55 (1), 4455. Petersen, L. R., & Roy, A. (1985). Religiosity, anxiety, and meaning and purpose: ReligionÂ’s consequences for psychological well-being. Review of Religious Research, 27 49-62.
230 Peterson, C., Seligman, M. E. P., & Vaillant, G. E. (1988). Pessimistic explanatory style is a risk f actor for physical illness: A thirty-five year longitudinal study. Journal of Personality and Social Psychology, 55 23-27. Piper, Watty. (1930). The little engine that could New York: Platt & Munk Publishers. Pitkala, K. H., Laakkonen, M. L., Strandberg, T. E., & Tilvis, R. S. (2004). Positive life orientation as a predictor of 10-year outcome in an aged population. Journal of Clinical Epidemiology, 57 409-414. Polk, L. (1997). Toward a middle range theory of resilience. Advances in Nursing Science, 19 (3), 1-13. Powell, L. H., Shahabi, L., & Thoresen, C. E. (2003). Religion and Spirituality, Linkages to Physical Health. American Psychologist, 58 (1), 36-52. Preacher, K. J., & Hayes, A. F. (2004) SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behavior Research Methods, Instruments, & Computers, 36 (4), 717-731. Preacher, K. J., & Leonardelli, G. J. ( 2003). Calculation for the Sobel Test: An interactive calculation tool for mediation tests. Retrieved July 1, 2006, from http://www.unc.edu/~preacher/sobel/sobel.htm Rabkin, J. G., Remien, R., Katoff, L., & Williams, J.B.W. (1993). Resilience in adversity among long-term survivors of AIDS. Hospital and Community Psychiatry, 44 (2), 162-167. Radloff, L. W. (1977). The CES-D Scal e: A self report depression scale for research in the general population. Applied Psychological Measurement, 1 385-401.
231 Radloff, L. W., & Teri, L. (1986). Use of the Center for Epidemiological Studies-Depression Scale with older adults. Clinical Gerontology, 5 119-137. Reker, G. T., Peacock, E. J., & Wong, P. T. (1987). Meaning and purpose in life and well-being: A life-span pers pective. Journal of Gerontology 42 44-49. Rew, L., Taylor-Seehafer, M., Thomas, N. Y., & Yockey, R. D. (2001). Correlates of resilience in homeless adolescents. Image: Journal of Nursing Scholarship, 33 (1), 33-41. Richardson, G. E. (2002). The metath eory of resilience and resiliency. Journal of Clinical Psychology, 58 307-321. Robert, S., & House, J. S. (1996). SES differentials in health by age and alternative indicators of SES. Journal of Aging and Health, 8 359-388. Roberts, A. (2003). In the eye of the stor m? Societal aging and the future of public-service reform. Public Administration Review, 63 (6), 720-733. Robinson-Whelen, S., Kim, C., MacCallum, R. C., & Kiecolt-Glaser, J. K. (1997). Distinguishing optimism from pessimism in older adults: Is it more important to be optimistic or not to be pessimistic? Journal of Personality and Social Psychology 73, 1345-1353. Rook, K. S. (1990). Stressful aspects of older adultsÂ’ social relationships: current theory and research. In M. A. P. St ephens, J. H. Crowther, S. E. Hobofoll, & D. L. Tennenbaum, (Eds.), Stress and coping in later life pp. 173-192. Washington, DC: Hemisphere. Roos, N. & Havens, B. (1991). Predictors of successful aging: A twelve-year study of Manitoba elderly. American Journal of Public Health, 81 (1), 63-68.
232 Rosenberg, M., & McCullough, B. (1981). Mattering: Inferred significance and mental health among adolescents. Research in Community & Mental Health, 2 163-182. Rosow, L. (1985). Status and role cha nge throught the life cycle. In R. H. Binstock & E. Shanas (Eds.), Handbook of aging and the social sciences (2nd ed.), pp. 6293. New York: Van Nostrand Reinhold. Ross, L., & Mirowsky, J. (1989). Explai ning the social patterns of depression: Control and problem solving or support and taking? Journal of Health and Social Behavior, 30 206-219. Rothermund, K., & Brandstdter, J. ( 2003). Depression in later life: Crosssequential patterns and possible determinants. Psychology and Aging, 18 (1), 80-90. Rowe, J. W., & Kahn, R. L. (1987). Human Aging: Usual and successful. Science, 237 (10), 143-149. Rowe, J. W., & Kahn, R. L. (1997). Successful aging. Gerontologist, 37 433440. Rowe, J. W., & Kahn, R. L. (1998). Successful Aging. The MacArthur Foundation Study New York, NY: Dell Publisher. Rudinger, G., & Thomae, H. (1990). The Bonn Longitudinal Study of Aging: Coping, life adjustment, and life satisfaction. In P. Baltes & M. Baltes (Eds.), Successful aging pp. 265-295. New York: Cambridge University Press. Rutter, M. (1985). Resilience in the f ace of adversity: Protective factors and resistance to psychiatric disorder. British Journal of Psychiatry, 147 598-611. Rutter, M. (1987). Psychosocial re silience and protective mechanisms. American Journal of Orthopsychiatry, 57 316-331.
233 Ryff, C. D. (1982). Successful aging: A developmental approach. The Gerontologist, 22 209-214. Ryff, C. D. (1989). In the eye of th e beholder: Views of psychological wellbeing among middle-aged and older adults. Psychology and Aging, 4 (2): 195-210. Salsman, J. M., Brown, T. L., Brechting, E. H., & Carlson, C. R. (2005). The link between religion and spirituality and psychologi cal adjustment: The mediating role of optimism and social support. Personality and Social Psychology Bulletin, 31 (4), 522-535. Salovey, P., Rothman, A. J., Detweiler, J. B., & Steward, W. T. (2000). Emotional states and physical health. American Psychologist, 55 (1), 110-121 SAS Institute, Inc. Cary, NC, 27513. Schaefer, J. A., & Moos, R. H. (1992). Li fe crises and personal growth. In B. N. Carpenter (Ed.), Personal coping: Theory, research, and application (pp. 149-170). Westport, CT: Praeger. Scheidt, R. J., Humphreys, D. R., & Yorgason, J. B. (1999). Successful aging: WhatÂ’s not to like? The Journal of Applied Gerontology, 18 277-282. Scheier, M. R., & Carver, C. S. (1985). Optimism, coping, and health: Assessment and implications of generalized outcome expectancies. Health Psychology, 4 219-247. Scheier, M. F., Carver, C.S. (1992). Effects of optimism on psychological and physical well-being: theoretical overview and empirical update. Cognitive Therapy Research, 16 201-228.
234 Scheier, M. F., Carver, C.S., & Bridges, M. W. (1994). Distinguishing optimism from neuroticism (and trait anxiety, self-maste ry, and self-esteem): A reevaluation of the Life Orientation Test. Journal of Personality and Social Psychology, 67 (6), 1063. Schoenfeld, D. E., Malmrose, L. C., Blazer, D. G., Gold, D. T., & Seeman, T. E. (1994). Self-rated health and mortality in th e high-functioning elderl y Â– A closer look at healthy individuals: MacArthur fi eld Study of Successful Aging. Journal of Gerontology: Medical Sciences 49 (3), M109-M115. Schulz, R., Bookwala, J., Knapp, J. E., Sc heier, M., & Williamson, G. M. (1996). Pessimism, age, and cancer mortality Psychology and Aging, 11 304-309. Schulz, R., & Heckhausen, J. (1996). A life span model of successful aging. The American Psychologist, 51 702-714. Scott, W. K., Macera, C. A., Cornman, C. B., & Sharpe, P. A. (1997). Functional health status as a predictor of mortality in men and women over 65. Journal of Clinical Epidemiology, 50 291-296. Seeman, T. E., Berkman, L. F., Charpentie r, P. A., Blazer, D. G., Albert, M. S. & Tinetti, M. E. (1995). Beha vioral and psychosocial predicto rs of physical performance: MacArthur studies of successful aging. Journal of Gerontology, 50A M177-M183. Seeman, T. E., Bruce, M. L. & McAvay, G. J. (1996). Social network characteristics and onset of ADL disability : MacArthur Studies of Successful Aging. Journal of Gerontology: Social Sciences, 51B (4), S191-S200. Seeman, T. E., Dubin, L. F., & Seeman, M. (2003). Religiosity/spirituality and health: A critical revi ew of the evidence for biological pathways. American Psychologist, 58 53-63.
235 Selye, H. (1956). The stress of life New York: McGraw-Hill. Shnek, Z. M., Irvine, J., Stewart, D ., & Abbey, S. (2001). Psychological factors and depressive symptoms in ischemic heart disease. Health Psychology, 20 141-145. Shevlin, M., Brunsden, V., & Miles, J. N. V. (1998). Satisfaction with Life Scale: Analysis of factorial invarian ce, mean structure, and reliability. Personality and Individual Differences, 25 911-916. Shin, D. C., & Johnson, D. M. (1978). Avowed happiness as an overall assessment of the quality of life. Social Indicators Research, 5 475-492. Shrout, P. E., & Bolger, N. (2002) Mediation in experimental and nonexperimental studies: New pr ocedures and recommendations. Psychological Methods, 7 422-445. Shrout, P. E., & Yager, T. J. (1989). Re liability and validity of screening scales: Effect of reducing scale length. Journal of Clinical Epidemiology, 42 69-78. Siegler, I. C., & Brummett, B. H. ( 2000). Associations among NEO personality assessments and well-being at midlife: Facet-level analyses. Psychology and Aging, 15 710-714. Skevington, S. M. (1993). Depression a nd causal attributions in the early stages of a chronic painful disease: A longitudinal study of early synovitis. Psychology and Health, 8 51-64. Snowdon, D. A., Ostwald, S. K., & Kane, R. I. (1989). Education survival and independence in elderly cat holic sisters, 1936-1988. American Journal of Epidemiology, 130 999-1012.
236 Sobel, M. E., (1982). Asymptotic confid ence intervals for indirect effects in structural equation models. In S. Leinhardt (Ed.), Sociological Methodology 1982 (pp. 290-312). San Francisco: Jossey-Bass. Solomon, Z. & Prager, E. (1992). Elde rly Israeli Holocaust survivors during the Persian Gulf War: A study of psychological distress. American Journal of Psychiatry, 149 (12), 1707-1710. Spector, W. D., Katz, S., Murphy, J. B., & Fulton, J. P. (1987). Hierarchical relationship between activities of daily living and instrumental activities of daily living. Journal of Chronic Diseases, 40 (6), 481-489. Sperry, L. (1992). Recent developments in neuroscience, behavior medicine, and psychoneuroimmunology: Implications for physical and psychological well-being. Individual Psychology Journal of Adle rian Theory, Research and Practice, 48 480-487. Staudinger, U. M., & Fleeson, W. (1996). Self and personality in old and very old age: A sample case of resilience? Development & Psychopathology, 8 867-885. Staudinger, U. M., Marsiske, M., & Baltes, P. B. (1993). Resilience and levels of reserve capacity in later adulthood: Perspectives from life-span theory. Development and Psychopathology, 5 541-566. Staudinger, U. M., Marsiske, M., & Baltes, P. B. (1995). Resilience and reserve capacity in later adulthood: Potentials and li mits of development across the life-span. In D. Cicchetti, & D. J. Cohen (Eds.), Developmental psychopathology. Risk, Disorder and Adaptation, vol. 2 (pp. 801-847). New York: Wiley and Sons.
237 Steele, A., & Wade, T. D. (2004). The contribution of optimism and quality of life to depression in an acute coronary syndrome population. European Journal of Cardiovascular Nursing, 3 231-237. Strawbridge, W. J., Cohen, R. D., Shem a, S. J., & Kaplan, G. A. (1996). Successful aging: Predictors and associated activities. American Journal of Epidemiology, 144 135-144. Strawbridge, W. J., Kaplan, G. A., Cam acho, T., & Cohen, R. D. (1992). The dynamics of disability and functional change in an elderly cohort: Results from the Alameda County Study. Journal of American Geriatrics Society, 40 799-806. Strawbridge, W. J., Wallhagen, M. I., & Cohen, R. D. (2002). Successful aging and well-being: Self-rated compared with Rowe and Kahn. The Gerontologist, 42 (6), 727-733. Steverink, N., Lindenberg, S., & Ormel, J. (1998). Towards understanding successful ageing: Patterned ch ange in resources and goals. Ageing and Society 18 441-467. Steverink, N., Westerhof, G. J., B ode, C., & Dittmann-Kohli, F. (2001). The personal experience of aging, individual resources, and subjective well-being. Journal of Gerontology: PYSCHOLOGICAL SCIENCES, 56B (6), P364-P373. Suchman, E., Phillips, S. B., & Streib, G. (1958). An analysis of the validity of health questionnaires. Social Forces, 36 223-232. Swartz, E. I. (1978). The older adul t: Creative use of leisure time. The Journal of Geriatric Psychiatry, 11 85-87.
238 Syme, S. L. (2003). Psychosocial interv entions to improve successful aging. Annals of Internal Medicine, 139 400-402. Symister, P., & Friend, R. (2003). The infl uence of social support and problematic support on optimism and depression in chronic illness: A prospective study evaluating self-esteem as a mediator. Health Psychology, 22 (2), 123-129. Taylor, S. E., & Armor, D. A. (1996) Positive illusions and coping with adversity. Journal of Personality, 64 873-898. Taylor, S. E., & Brown, J. D. (1994). Positive illusions and we ll-being revisited: Separating fact from fiction. Psychological Bulletin, 116 21-27. Taylor, S. E., Kemeny, M. E., Aspinwall, L. G., Schneider, S. G., Rodriguez, R. & Herbert, M. (1992). Optimism, coping, psychological distress, and high-risk sexual behavior among men at risk for Acquire d Immunodeficiency Syndrome (AIDS). Journal of Personality and Social Psychology, 63 460-473. Taylor, S. E., Kemeny, M. E., Reed, G. M., Bower, J. E., & Gruenewald, T. L. (2000). Psychological resources, positive illusions, and health. American Psychologist, 55 (1), 99-109. Tellegen, A., Lykken, D., Bouchard, T. J., Wilcox, K. J., Segal, N. J., Rich, S. (1988). Personality similarity in twins reared apart and together. Journal of Personality and Social Psychology, 54 1031-1039. Terrill, D. R., Friedman, D. G., Gottschalk, L. A., & Haaga, D. A. F. (2002). Construct validity of the Life Orientation Test. Journal of Personality Assessement, 79 (3), 550-563.
239 Tessler, R., & Mechanic, D. (1978). Ps ychological distress and perceived health status. Journal of Health and Social Behavior, 19 254-262. Thoits, P. A., & Hewitt, L. N. ( 2001). Volunteer work and well-being. Journal of Health and Social Behavior, 42 (June), 115-131. Thoresen, C. E. (1999). Spirituality a nd health: Is there a relationship? Journal of Health Psychology, 4 291-300. Tiger, L. (1999). The past of an illusion: How optimism brings us to God. Free Inquiry 19, 28-29. Tissue, T. (1972). Another look at self-rated health among the elderly. Journal of Gerontology, 27 91-94. Tilvis, R. S., Pitkala, K. H., Jolkkonen, J., & Strandberg, T. E. (2000). Social networks and dementia. Lancet, 356 77-78. Tobin, S., & Neugarten, B. (1961). Life satisfaction and social interaction in the aging. Journal of Gerontology, 16 344-46. Turvey, C. L., Wallace, R. B., & Herzog, R. (1999). A revised CES-D measure of depressive symptoms and a DSM-based m easure of major depressive episodes in the Elderly. International Psychogeriatrics, 11 (2), 139-148. Ulbrich, P. M., & Warheit, G. J. (1989). Social support, stress, and psychological distress among older Black and White adults. Journal of Aging and Health, 1 286-305. Umberson, D., House, J. S., Chen, M. D ., Hopkins, K., & Slaten, E. (1996). The effect of social relationships on psychological well-being: Are men and women really so different? American Sociological Review, 61 837-857.
240 Unger, J. B., Johnson, C. A., & Marks, G. (1997). Functional decline in the elderly: Evidence for direct and stress-buffe ring protective effects of social interactions and physical activity. Annals of Behavior Medicine 19, 152-160. Unger, J. B., McAvay, G., Bruce, M. L., Berkman, L., & Seeman, T. (1999). Variation in the impact of social networ k characteristics on physical functioning in elderly persons: MacArthur St udies of Successful Aging. Journal of Gerontology: Social Sciences, 45B( 5), S245-S251. U. S. Bureau of the Census. (2000). Statistical Abstract of the United States Washington, D.C.: U. S. Department of Commerce. Vickers, K. S., & Vogeltanz, N., E. (2000) Dispositional optimism as a predictor of depressive symptoms over time. Personality and Individual Differences, 28 259-272. Vogt, T. M., Mullooly, J. P., Ernst, D., Pope C. R., & Hollis, J. F. (1992). Social networks as predictors of ischemic hear t disease, cancer, stroke, and hypertension: Incidence, survival and mortality. Journal of Clinical Epidemiology, 45 659-666. Wagnild, G., & Young, H.M. (1990). Resilience among older women. Image: Journal of Nursing Scholarship, 22 (4), 252-255. Wagnild, G., & Young, J.M. (1993). De velopment and psychometric evaluation of the resilience scale. Journal of Nursing Measurement, 1 (2), 165-178. Wallace, K. A., Bisconti, T. L., & Bergeman, C.S. (2001). The mediational effect of hardiness on social support and optimal outcomes in later life. Basic and Applied Social Psychology, 23 (4), 267-279.
241 Walter-Ginzburg, A., Shmotkin, D., Bl umstein, T., & Shorek, A. (2005). A gender-based dynamic multidimensional longitudinal analysis of resilience and mortality in the old-old in Israel: The cross-sectional and long itudinal aging study (CALAS). Social Science & Medicine, 60 1705-1715. Ware, J. E. Jr. & Sherbourne, C. D. (1992). The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Medical Care, 30 473-483. Watson, D. (1988b). The vicissitudes of mood measurement: Effects of varying descriptors, time frames, and response formats on measures of positive and negative affects. Journal of Personality and Social Psychology, 55 128-141. Watson, D. (1988a). Intraindividual and interindividual analyses of Positive and Negative Affect: Their relation to health complaints, perceived stress, and daily activities. Journal of Personality and Social Psychology, 54 1020-1030. Watson, D., & Clark, L. A. (1984). Ne gative Affectivity: The disposition to experience aversive emotional states. Psychological Bulletin, 96 465-490. Watson, D., Clark, L. A., & Carey, G. (1988). Positive and negative affectivity and their relation to anxiety and depressive disorders. Journal of Abnormal Psychology, 97 346-353. Watson, D. L., Clark, A., & Tellegen, A. (1984). Cross-cultural convergence in the structure of mood: A Japanese replic ation and a comparison with U.S. findings. Journal of Personality and Social Psychology, 47 127-44.
242 Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of Brief Measures of Positive and Ne gative Affect: The PANAS Scales. Journal of Personality and Social Psychology, 54 (6), 1063-1070. Waxman, H. M., Carner, E. A., & Klein, M. (1984). Underutilization of mental health professionals by community elderly. The Gerontologist, 24 23-30. Werner, E. E. & Smith, R. S. (1977). Ka uaiÂ’s children come of age. Honolulu: University of Hawaii Press. Wethington, E., & Kessler, R. C. (1986). Perceived support, received support, and adjustment to stressful life events. Journal of Health and Social Behavior, 27 78-89. Wetle, T. (1997). Living longer, aging betterÂ—aging research comes of age. Journal of the American Medical Association, 278 1376-77. Wilkinson, A. M., & Lynn, J. (2001). The end of life. In R. H. Binstock & L. K. George (Eds.), Handbook of Aging and the Social Sciences, (5th ed Pp. 444-461). San Diego: Academic Press. Willis, T. (1986). Stress and coping in early adolescence: Relationships to substance use in urban school samples. Health Psychology, 5 503-529. Wilson, W. (1967). Correlat es of avowed happiness. Psychological Bulletin, 67 294-306. Wink, P., Dillon, M., & Larsen, B. (2005) Religion as moderator of depressionhealth connection. Research on Aging, 27 (2), 197-220. Witter, R. A., Stock, W. A., Okun, M. A., & Haring, M. J. (1985). Religion and subjective well-being in adult hood: A quantitative synthesis. Review of Religious Research, 26 332-342.
243 Wolinsky, F. D. & Johnson, R. J. (1992). Perceived health status and mortality among older men and women. Journal of Gerontology, 47 (6), S304-S312. Woodward, K. L. (1997). Is God listening? Newsweek, 31 (3), 57-64. Woodard, M. (1987). Voluntary association membership among Black Americans. Sociology Quarterly, 28 285-301. Yardley, J. K., & Rice, R. W. (199 1). The relationship between mood and subjective well-being. Social Indicators Research, 24 101-111. Zimmer, Z., Hickey, T., & Searle, M. (1995). Activity part icipation and wellbeing among older people with arthritis. The Gerontologist, 35 (4), 463-471. Zheng, D., Macera, C. A., Croft, J. B., G iles, W. H., Davis, D., & Scott, W. K. (1997). Major depression and all-cause mo rtality among White adults in the United States. American Journal of Epidemiology, 7 213-218.
About the Author Teri A. Trede received a BachelorÂ’s Degree in Criminal Justice from the University of South Florida in 1980. Upon gra duation, she worked for a number of years in the domestic violence, substance abuse and mental health fields. She returned to USF and obtained a MasterÂ’s Degree in Public He alth in 1999. Following the completion of her MasterÂ’s Degree, she enrolled in the Ph.D. program in the College of Public Health at USF. Throughout her doctoral program, Ms. Trede taught undergraduate courses for USF as well as worked as a field interviewer for the Elderly Research Center of Case Western Reserve University. Ms. Trede obtained her Ph.D in 2006. Her research interests include successful aging, resilience, religiosity, volun teerism, sexual health and decision making, and consumer health issues.