Social determinants of health among older adults: evidence from the utah fertility, longevity, and aging (flag) study

For some decades, social relationship has been a central theme in research on health and wellbeing. The literature documents two separate but related components of social relationship-social network and social support-both of which are believed to impact health independent of the other. Using data from the Utah Fertility, Longevity, and Aging (FLAG) study, the current study investigated the associations of dimensions of social connectedness (network and satisfaction with network) and perceived social support (affective, confidant, and instrumental support) to physical and mental health, and examined whether or not the association between social connectedness and physical and mental health of older adults was attributable to perceived social support. Results of the study showed the dimensions of social connectedness (network, and satisfaction with network) and perceived social support (affective, confidant, and instrumental support) were positively correlated. These dimensions, with the exception of the network dimension, were also positively associated with physical and mental health. Independent samples t-test showed individuals who obtained higher scores on the satisfaction with network dimension, and affective, confidant, and instrumental support dimensions were more likely to have higher physical and mental health scores than those who obtained lower scores on these dimensions. Logistic regression analyses showed high scores on affective and instrumental support were associated with higher odds of reporting good physical health. Similarly, high scores on the satisfaction with network dimension were associated with higher odds of reporting good mental health. Hierarchical multiple regression analyses showed affective and instrumental support, and satisfaction with network dimension were significant predictors of physical and mental health when the effects of covariates were controlled for. Results of moderation analyses showed significant conditional effects of social connectedness and perceived social support on physical and mental health. The interaction term (Connectedness_X_Support) was not significant. Perceived social support did not moderate the relationship between social connectedness and physical and mental health. Other correlates of physical and mental health included age, gender, and socio-economic status (SES). An increase in age corresponded with favorable mental health. Higher SES was associated with reporting good physical and mental health. Being female was associated with greater likelihood of reporting poor physical and mental health. Findings generally suggest social connectedness and perceived social support may affect different aspects of health independent of the other. Findings also suggest perceived social support may be relatively more important to the health and wellbeing of older adults than social connectedness and underscore the relative importance older adults attach to quality rather than quantity of social ties. Implications for social work practice and education, policy, and research are discussed.

SOCIAL DETERMINANTS OF HEALTH AMONG OLDER ADULTS: EVIDENCE FROM THE UTAH FERTILITY, LONGEVITY, AND AGING (FLAG) STUDY by Samuel Asante A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy College of Social Work The University of Utah August 2015 Copyright © Samuel Asante 2015 All Rights Reserved The University of Utah Graduate School STATEMENT OF DISSERTATION APPROVAL The dissertation of Samuel Asante has been approved by the following supervisory committee members: Marilyn Luptak , Co-Chair 06/11/2015 Date Approved Frances Wilby , Co-Chair 06/11/2015 Date Approved Jason Castillo , Member 06/11/2015 Date Approved Ken Smith , Member 06/11/2015 Date Approved Aster Tecle , Member 06/11/2015 Date Approved and by Lawrence Henry Liese , Chair/Dean of the Department/College/School of Social Work and by David B. Kieda, Dean of The Graduate School. ABSTRACT For some decades, social relationship has been a central theme in research on health and wellbeing. The literature documents two separate but related components of social relationship-social network and social support-both of which are believed to impact health independent of the other. Using data from the Utah Fertility, Longevity, and Aging (FLAG) study, the current study investigated the associations of dimensions of social connectedness (network and satisfaction with network) and perceived social support (affective, confidant, and instrumental support) to physical and mental health, and examined whether or not the association between social connectedness and physical and mental health of older adults was attributable to perceived social support. Results of the study showed the dimensions of social connectedness (network, and satisfaction with network) and perceived social support (affective, confidant, and instrumental support) were positively correlated. These dimensions, with the exception of the network dimension, were also positively associated with physical and mental health. Independent samples t-test showed individuals who obtained higher scores on the satisfaction with network dimension, and affective, confidant, and instrumental support dimensions were more likely to have higher physical and mental health scores than those who obtained lower scores on these dimensions. Logistic regression analyses showed high scores on affective and instrumental support were associated with higher odds of reporting good physical health. Similarly, high scores on the satisfaction with iv network dimension were associated with higher odds of reporting good mental health. Hierarchical multiple regression analyses showed affective and instrumental support, and satisfaction with network dimension were significant predictors of physical and mental health when the effects of covariates were controlled for. Results of moderation analyses showed significant conditional effects of social connectedness and perceived social support on physical and mental health. The interaction term (Connectedness_X_Support) was not significant. Perceived social support did not moderate the relationship between social connectedness and physical and mental health. Other correlates of physical and mental health included age, gender, and socio-economic status (SES). An increase in age corresponded with favorable mental health. Higher SES was associated with reporting good physical and mental health. Being female was associated with greater likelihood of reporting poor physical and mental health. Findings generally suggest social connectedness and perceived social support may affect different aspects of health independent of the other. Findings also suggest perceived social support may be relatively more important to the health and wellbeing of older adults than social connectedness and underscore the relative importance older adults attach to quality rather than quantity of social ties. Implications for social work practice and education, policy, and research are discussed. TABLE OF CONTENTS ABSTRACT ……………………………………………………………………………. iii LIST OF TABLES .……………………………………………………..………………vii LIST OF FIGURES …………………………………………………………………….. ix AKNOWLEDGEMENTS …………………………………………………………….... x CHAPTERS 1. INTRODUCTION…………………………………………………………………….1 Purpose of study .......................................................................................................5 Research questions and hypotheses .........................................................................6 Organization of study ...............................................................................................7 2. LITERATURE REVIEW……………………………………………………………..9 The aging of the population .....................................................................................9 Social relationships and health of older adults ......................................................11 Theoretical framework ...........................................................................................27 Theoretical and methodological issues in social relationship and health studies ..........................................................................................................38 3. RESEARCH METHODS ...…………………………………………………………44 Fertility, Longevity, and Aging (FLAG) study ......................................................44 Current study ..........................................................................................................46 4. FINDINGS…………………………………………………………………………...58 Descriptive data .....................................................................................................58 Social connectedness, perceived social support, and health ..................................69 Summary of results ................................................................................................84 5. DISCUSSION………………………………………………………………………..86 Social connectedness, perceived social support, and health: The association .......86 vi What dimensions of social connectedness and perceived social support are important to physical and mental health? ..............................................................89 Variations in association of social connectedness and perceived social support to physical and mental health ....................................................................94 The moderation effect of perceived social support ................................................95 Social connectedness, perceived social support, and socio-demographic characteristics .........................................................................................................97 What socio-demographic characteristics are important to physical and mental health ...................................................................................................98 Integrative summary-strengths, limitations, and implications of study ............101 Summary………………………………………………………………………..107 Appendices A: STUDY INSTRUMENTS ..................................................................109 B: CONSENT LETTER: CONSENT AND AUTHORIZATION DOCUMENT ......................................................................................121 REFERENCES ...................................................................................................131 LIST OF TABLES Table Page 1. Summary statistics for dimensions of social connectedness, perceived social support, and health measures ...............................................................................51 2. Socio-demographic characteristics of study participants ......................................59 3. Mean scores of social connectedness, perceived social support, and health measures ................................................................................................................61 4. X2-test - Distribution of sample demographic characteristics according to level of social connectedness ............................................................................63 5. X2-test - Sample demographic characteristics and perceived social support .......64 6. Means score differences in dimensions of social connectedness in relation to physical and mental health (t-test) .........................................................................66 7. Variations in dimensions of perceived social support in relation to physical and mental health (t-test) .......................................................................................68 8. Correlations among study variables ......................................................................70 9. Logistic regression: Predicted probabilities of good physical health ...................74 10. Logistic regression: Predictors of good mental health ..........................................75 11. Co-efficients and standard errors from regression of physical health scores on covariate and predictor variables ..........................................................77 12. Regression of mental health scores on covariate and predictor variables…….....79viii 13. Moderation analysis: Effect of social support on relationship between social connectedness and physical health.........................................................................82 14. The moderation effect of social support on relationship between social connectedness and mental health ..........................................................................83 LIST OF FIGURES Figure Page 1. Social relationship and health model ....................................................................24 2. Network, support, and health model .....................................................................35 ACKNOWLEDGEMENTS "Trust in the LORD with all thine heart; and lean not unto thine own understanding. In all thy ways acknowledge Him, and He shall direct thy paths," Proverbs 3:5-6. I gratefully acknowledge the following individuals and organizations for their assistance and support: The Utah Fertility, Longevity, and Aging (FLAG) Study research group by whose effort I obtained data for this study; each member of the dissertation committee, for providing helpful guidance throughout the research process and for enriching this study with personal insight; Special gratitude to Frances Wilby, PhD., and Marilyn Luptak, PhD., my dissertation Co-chairs, who inspired me with their commitment to my successful completion of the doctoral program; Jason Castillo, PhD., dissertation committee member, who has been my right arm for half of a decade, and provided guidance and immeasurable support throughout the research process; Ken R. Smith, PhD., for granting permission to access and use the FLAG data; Aster Tecle, PhD., who had a personal interest in my wellbeing and provided an invaluable support throughout this project. I would like to thank Amanda S. Barusch, PhD., who created the path and sustained my interest in aging research, Brad W. Lundahl, PhD., and Ms. Mirela Rankovic, for their unwavering support and encouragement throughout this project. A special thank-you goes to Dr. & Mrs. Michael Adjei-Poku, and Ms. Georgina xi Tuffour, for their support and for keeping me on track by consistently enquiring about my progress in the research process. I would also like to say a loving thank-you to my family back in Ghana. A special thank-you goes to my Mother, Mrs. Lucy Asante, who saw the potential in me, trusted in my ability to excel in every endeavor, and sacrificed all she had to put me through school. I love you, Mother. To all the teaching and nonteaching staff in the College of Social Work, University of Utah, and members of the Central SDA Church, Salt Lake City, Utah, whose names could not be captured here, I say thank you. God Bless! CHAPTER 1 INTRODUCTION In the next few decades, the U.S. will experience a transformation in the demographic structure, with the proportion of older adults, 65 years and older, projected to outnumber those younger than 18 years by 2060 (US Census Bureau, 2013). In 2011, the U.S. Census Bureau estimated there were 41.4 million persons aged 65 and older, which represented 13% of the national population. By 2030, this number is expected to increase to more than 72 million and, by 2050, more than double to 88 million, with the more frail (85 years and older) projected to quadruple to 19 million (Administration on Aging (AoA), 2013). The healthy aging of the population, from the medical standpoint, is seen as the result of numerous factors including improvement in health and medicine (Perkins, Multhaup, Perkins, & Barton, 2008). From a social viewpoint, however, scholars contend that productive and healthy aging is the result of active integration and participation of older adults in society, two important conditions made possible through social relationships (British Columbia Ministry of Health (BCMH), 2004; Lennartsson & Silverstein, 2001; Zunzunegui, Alvarado, Del Ser, & Otero, 2003). Erikson and colleagues' (1986) classical work emphasized that successful aging and healthy development in late life involves reflection and renewal of previous life balances around "themes of hope, purpose, competence, 2 commitment, love and care" (pp. 55-56). Older persons achieve these thematic renewals by their engagement with people, institutions, organizations, and relationships that in the present life, constitute their world, and by reexamining earlier life commitments, interactions, and relationships. Social relationships are fundamental to human survival, and are significantly involved in the attainment and maintenance of good health and wellbeing (Ashida & Heaney, 2008; Steptoe, Shankar, Demakakos, & Wardle, 2013). Social relationship has been variously defined and measured diversely across studies and disciplines. Regardless of the differences, however, two major components of social relationships have consistently been studied and documented. These include social network, and social support (Antonucci, Birditt, & Ajrouch, 2011; Antonucci, Birditt, & Akiyama 2009; Fiori, Antonucci, & Cortina, 2006; Holt-Lunstad, Smith, & Layton, 2010). These components, also considered as the structural and functional characteristics of social relationships, have been linked to mental health (Fiori et al., 2006), physical health morbidity (DiMatteo, 2004; Perkins, Ball, Kemp, & Hollingsworth, 2013), and mortality (Antonucci, Birditt, & Webster, 2010; Cornwell & Waite, 2009; Holt-Lundstad et al., 2010). Social relationships are considered important for older adults' physical health and psychological wellbeing (Choi & McDougall, 2009; Fiori et al., 2006; Steptoe et al., 2013). Strong ties with families and friends have been found to improve mental and physical health, positively influence health behaviors, reduce mortality risk (BCMH, 2004; Chen, Hicks, & While, 2013; DiMatteo, 2004; Steptoe et al., 2013; Uchino, 2013; Umberson & Karas, 2010), and enable older adults to stay in the community rather than 3 being institutionalized (Aschbrenner, Mueser, Bartels, & Pratt, 2011). Additionally, supportive relationships have been linked to the provision of emotional security (Fiori et al., 2006). With its absence often experienced as emotional (loneliness) and social isolation, older adults appraise their social relationships on the basis of the degree to which they feel connected and supported (Ashida & Heaney, 2008; BCMH, 2004; Cornwell & Waite, 2009; Golden et al., 2009; McPherson, Smith-Lovin, & Brashears, 2006; Steptoe et al., 2013). Social connectedness and social support have not always been considered separately in previous studies. This is partly the result of their linear relationship, with social support being a function of social relations that is provided by members in one's social network. In most studies, for instance, having a companion was synonymous with social support (Aboim, Vasconcelos, & Wall, 2013; Hawkley, Masi, Berry, & Cacioppo, 2006; Kroenke, Kubzansky, Schernhammer, Holmes, & Kawachi, 2006; Pedersen, Andersen, & Curtis, 2012; Yuan et al., 2011) regardless of whether or not support was provided. Again studies examining isolation and loneliness have to a large extent been conducted in the context of social support (Chen et al., 2013; Dykstra, & Fokkema, 2007; Liu & Guo, 2007; Tomaka, Thompson, & Palacios, 2006) where availability of social support indicated the presence of social relations or ties and thus the absence of loneliness feelings. Some studies, however, suggest that availability of companionship does not guarantee that social support will actually be provided (Antonucci et al., 2009; Ashida & Heaney, 2008; Nurullah, 2012). It is important to note that not all social relationships involve the exchange of support (Antonucci et al., 2009). To be clear, individuals can feel 4 socially disconnected or isolated and unsupported while surrounded by a multitude of potential support providers. A few studies on social relationships have examined the influence of isolated aspects of social relationships such as total level of connectedness and amount of social support on health and wellbeing of older adults (Antonucci, 2009; Broadhead et al., 1998; Kahn, 1979; Wong, Yoo, Stewart, 2005). While this method is important and enlightening, there are theoretical and empirical reasons to suspect that adding up the individual aspects of relationships to create a unidimensional construct (level of connectedness and perceived social support) does not compare the effect of being nested in a relationship with particular set of characteristics (e.g., frequent interaction with family and friends or receiving emotional support). In other words, by examining social connectedness and social support as singular, nondimensional constructs, it becomes difficult to distinctly identify the dimensions within each construct and their health implications, particularly among older adults for whom the importance of social relationships cannot be underestimated. In spite of the empirical evidence linking some of these dimensions of connectedness and social support to health (Alpass & Neville, 2003; Chen et al., 2013; Hsu, 2007; Moon, Park & Cho, 2010; Tay, Tan, Diener, & Gonzalez, 2013), a limited number of studies exists that simultaneously examines the dimensions of social connectedness and social support and their relationship with health, thus making it difficult to draw a firm conclusion on the health implications of dimensions of social connectedness and social support. It, therefore, may be more informative to examine some of these dimensions and their association to physical health and mental wellbeing 5 of older adults. This study aims to investigate the association of dimensions of social connectedness (network and satisfaction with network) and perceived social support (affective, confidant, and instrumental) in relation to physical and mental health. As previously indicated, research on social relationships and health has focused on both structural (e.g., network-connectedness) and functional (e.g., social support) characteristics of social relationships. The structural characteristics, however, have received more attention compared to the functional characteristics. Few of these studies have examined the mechanisms by which social relationship and health are related. Given that the functional characteristics have generally been found to have greater impact on health than the structural characteristics (Besser & Priel, 2008; Teo, Choi, & Valenstein, 2013), it is important to investigate the influence of the major functional characteristic of social relationships which might be the singular, most important underlying mechanism through which the structural characteristics of social relationships and health are related: namely perceived social support. Purpose of study This study aimed to (1) investigate the association of dimensions of social connectedness (network and satisfaction with network) and perceived social support (affective, confidant, and instrumental) in relation to physical and mental health; and (2) to determine whether or not the association between social connectedness and physical and mental health of older adults is attributable to perceived social support. The study employed a quantitative design, utilizing secondary data from the longitudinal Utah Fertility, Longevity, and Aging (FLAG) study. Standardized measures included the 6 Medical Outcome Study Short Form 36 (SF 36), which examines functioning and wellbeing in older adults (McHorney, Ware, & Raczek, 1993), the Duke-UNC Functional Social Support Questionnaire (DUNCFSSQ), which measures an individual's perception of the amount and type of social support (Broadhead et al., 1998), and the Duke Social Support Index (DSSI), which measures the degree of a person's connectedness with others (Landerman, Georage, Campbell, & Blazer, 1989). The results may inform social work practice, education, policy, and research. Findings could lead to development of practice and policy interventions intended to increase social support and improve social ties through which support is given and received. Findings could also direct future research towards finding positive contributions older adults might make toward society (through which they would stay connected and supported) rather than focusing on their support needs and their demands on service provision. Research questions and hypotheses This study addressed the following research questions and hypotheses:  (Q1) Are there associations between the dimensions of social connectedness, perceived social support, and physical and mental health of older adults? Hypothesis 1: Dimensions of social connectedness (network and satisfaction with network) and perceived social support (affective, confidant, and instrumental support) will be positively associated with physical and mental health of older adults. 7  (Q2) Are there differences in how the dimensions of social connectedness and perceived social support relate with the physical and mental health of older adults? Hypothesis 2: Compared to the dimensions of social connectedness, higher scores on the dimensions of perceived social support will correspond with self-rated high physical and mental health scores.  (Q3) What dimensions of social connectedness and social support are important to physical and mental health of older adults? Hypothesis 3: Compared to the dimensions of social connectedness, the dimensions of social support will be significantly stronger predictors of self-rated physical and mental health.  (Q4) Does perceived social support moderate the relationship between social connectedness and physical and mental health of older adults? Hypothesis 4: Perceived social support will moderate the relationship between social connectedness and physical and mental health of older adults. Organization of study This study is organized into five chapters. Chapter 1 presents the study background, and highlights the purpose of the study, research questions, and research hypotheses guiding the study. Chapter 2 reviews the literature and highlights previous studies and theories that provide the foundation for this study. Chapter 3 focuses on research methods, including study design, sample, data collection procedures, measures, and statistical analyses. Chapter 4 focuses on results and presentation of findings. Chapter 8 5 addresses the discussion and conclusion. The results are interpreted in light of previous studies and theories forming the foundation of the study, and implications for social work practice, research, and policy are discussed. CHAPTER 2 LITERATURE REVIEW This chapter reviews the literature on social relationship and health in the population under study. The theoretical foundation of the study is also discussed. The chapter ends with a discussion on theoretical and methodological issues commonly found in social relationship and health studies. The aging of the population Currently, older adults are the fastest growing population on earth (Population Division, DESA, United Nations, 2013). It is estimated that 605 million people (about 9% of the world's population), aged 60 years and older are currently living around the globe. This figure is projected to rise to 2 billion by 2050, representing 16% of the world's population (World Health Organization (WHO), 2013a). Although age offers a benchmark for categorizing one as older adult, it is important to note that the term older adult means different thing to different people and often varies by geographic location (Gavrilov & Heuveline, 2007). On the basis of life expectancy at birth, there is a huge divide between the Western industrialized societies and the less industrialized societies of the world. While the age limit is set at 60 or 65 years for most contemporary Western societies, many developing countries consider old age as a period occurring anywhere10 from the mid-40s to the 70s (Encyclopedia Britannica, 2013). Most international documents use the term older adult loosely to indicate an individual who is 60 years and older (WHO, 2013b). Much of the world's older population is now concentrated in the more industrialized regions of the world, with six countries (China, US, India, Japan, Germany, and Russian Federation) accounting for 54% of the total (Population Division, DESA, United Nations, 2013). In the U.S., for instance, the Census Bureau in 2011 estimated there were 41.4 million individuals, aged 65 and older. This number represented 13% of the national population. By 2030, this number is expected to increase to more than 72 million, representing 20% of the national population, and more than double to 88 million by 2050 (US Census Bureau, 2011). The trend in population concentration around the globe is expected to change in the next few decades with most of the older population living in less industrialized regions of the world (Shetty, 2012; WHO, 2013). Since mortality rates among females are lower than male rates at old age, women constitute a significant majority of the older population. Influenced by decreasing fertility rates and remarkable increases in life expectancy, the aging of the population will continue, and even accelerate (National Institute of Health (NIH), 2013; WHO, 2013b; United Nations Population Fund (UNFPA), 2013). From the health or medical standpoint, population aging, in part, reflects successes in the areas of medicine and technology, which have both added years to life and life to years (Perkins et al, 2008; Takahashi & Tokoro, 2002). From the social standpoint, scholars contend that productive and healthy aging is the result of active 11 engagement of older adults in the society, a condition made possible through social relationships (BCMH, 2004; Lennartsson & Silverstein, 2001; Zunzunegui, Alvarado, Del Ser, & Otero, 2003). Social relationships and health of older adults The first major work on social relationship dates back to the industrial revolution of the 19th century. New phenomena such as migration, individualization, changing family structure, and unemployment drove new research into human relationships by sociologists, economists, and philosophers. As society was transformed by the industrial revolution, relationships were considered to have the ability to hold or disintegrate society (Coser 1971, pp. 133-136, pp. 184-185). Human beings are social by nature. As social beings, we possess a need to belong, a characteristic that is foundational to our emotions, thoughts, and interpersonal behaviors. The need to belong comprises a general "desire to form and maintain at least a minimum quantity of lasting, positive and significant interpersonal relationships" (Baumeister & Leary, 1995, p. 497). While differences exist in individual's need for belongingness and the means through which the need is met, satisfying this need inevitably involves a continual, emotionally satisfying interaction with others in a stable context that allows individuals to express concerns for one another's welfare (Baumeister & Leary, 1995; Heinrich & Gullone, 2006). Social relationship, for decades, has been a central theme in research on health and wellbeing, and is often represented with indicators that vary within and across disciplines. Social and health scientists interested in social networks, an indicator of 12 social relationships, have examined the health benefits and health risks associated with both large and small social networks (Cacioppo, Fowler & Christakis, 2009; Christakis & Fowler, 2008; Cornwell & Waite, 2009; Fowler & Christakis, 2008). Similarly, researchers have investigated and documented the effects of participation in social activities on people's health and wellbeing (Hsu, 2007; Moon et al., 2010). Researchers from disciplines such as social work, sociology, and nursing, who are interested in social support networks, have also examined the association between social support and health, and the extent to which people evaluate the support they receive as beneficial or detrimental (Golden et al., 2009; Kirke, 2013; Stephens, Alpass, Towers, & Stevenson, 2011; Uchino, 2006). Scholars have examined the direct influence of relationships on the psychological states of people. In his classical analysis of suicide, for instance, Durkheim (1897, p. 212) indicated the significant role that relationships play in suicide occurrence in a population. Compared to those more socially integrated, people who were less socially integrated were more likely to commit suicide. This finding has been confirmed in several studies across major social and behavioral disciplines (Compton, Thompson, & Kaslow, 2005; Cutright & Fernquist, 2001). Three major components of social relationships have been identified in the literature: social networks (a measure of social connectedness), social support, and support satisfaction (Antonucci & Akiyama, 2002; Antonucci & Wong, 2010; Antonucci et al., 2009). Together these components help determine the extent to which social relationship is a resource or a risk factor to individual's health and wellbeing. 13 Social relationships are considered important for older adults' physical health and psychological wellbeing (Choi & McDougall, 2009; Fiori et al., 2006) and are frequently seen as indicators of successful and healthy aging (Agahi & Parker, 2008; Canbaz, Sunter, Dabak & Peksen, 2003). It is widely accepted that relationships often provide older adults with meaningful roles, larger social networks, and different kinds of support, which have been linked to improved physiological functioning, coping abilities, and health behaviors (Agahi & Parker, 2008; Fiori et al., 2006, Lennartsson & Silverstein, 2001). Social connectedness (social network) The idea that humans need relationships to survive and that relationships are critical to human development is not new. The works of developmental psychologists including Erikson (1950), Bowlby (1988), and Ainsworth (1989) clearly indicate the importance of social relationships as the driving force in human development. From infancy to late adulthood, individuals live within webs of social ties, which are often called social networks (Ashida & Heaney, 2008; Kahn, 1979). The concept of social network is used to describe a finite set of actors and the relationship between them (Kirke, 2013). It has consistently been used in research as a measure of how connected one is to the social environment (Cornwell & Waite, 2009). Other indicators or dimensions of connectedness reported in the literature include frequency of interaction among network members and engagement in social activities (Cornwell & Waite, 2009; Lennartsson & Silverstein, 2001). Social networks can vary enormously in size, type, and pattern and benefits or 14 resources one may obtain (Thoit, 1982; Cohen & Wills, 1985). They are subject to change over time as new ties are formed or broken (Kirke, 2013; Shaw, Krause, Liang, & Bennett, 2007). Social networks are typically grouped into two categories: formal and informal (Kirke, 2013). Formal network involves one's association to formal organization such as a health care agency. Informal network involves family ties (e.g., spouse, children, and siblings) and friendship ties (often involving association with friends, and neighbors) (Clutier-Fisher, Kobayashi, Hogg-Jackson, & Roth, 2006). Although these ties are sometimes considered a source of psychological distress by exerting excessive demands on the individual, belonging to a healthy social network makes people feel respected, useful, cared for, loved, and cherished (Birditt, Jackey, & Antonucci, 2009; Gurung, Taylor, & Seeman, 2003). This has a strong protective effect on physical health and psychological wellbeing (WHO, 2003). The absence of social network is often experienced in the form of social isolation and emotional isolation (loneliness) (Victor, Scambler, Bond, & Bowling, 2000). As in all age groups, maintaining large and supportive social networks is important for older adults. From a combined standpoint of biological (e.g., simple deterioration theory) and social (e.g., activity theory) theories aging typically involves profound challenges to remaining socially connected (Bengtson, Gan, Putney, & Silverstein, 2009, pp. 31-32; Goldsmith, 2012). While the decrease in ability to form new relationship obviously leads to a decrease in social contact, research has shown that aging is marked by a renewal, maintenance, and formation of new and meaningful relationships (Antonucci et al., 2009; Kahn, 1979; Marjolein, Hoogendijk, & van Tilburg, 2013). Researchers have contended with the idea that social isolation is a normal aspect 15 of aging, and that loss of ties is characteristic of old age. Findings, however, are mixed. While some studies report a negative association between age and properties of network (size, and frequency of interaction), others indicate a positive relationship between these elements (Shaw et al., 2007). These findings are incongruous with the widely held view that aging generally has a negative influence on social ties (Cornwell, 2008). Research has shown that older adults who maintain large and supportive networks are often those who live with others, at least with a spouse (Wong, 2011). Although there are instances where older adults live alone, it is often argued that such adults tend to have large networks due to their perceived need for interaction and constant need of support (Schroot, Fernandez-Ballesteros, & Rudinger, 1999). Large and supportive networks ensure frequent contact with others through regular participation in social activities (Perkins et al., 2008). Some studies have also shown that greater sense of belongingness and lower levels of isolation and loneliness among older adults are indicative of larger proximate networks characterized by more intensive support exchanges (Ashida & Heaney, 2008; Golden et al., 2009; Kobayashi, Cloutier-Fisher, & Roth, 2009; Schroot et al., 1999). Older adults with meaningful connections report that involvement with others enhances self-image, and contributes to a positive self-attitude and self-acceptance (Reichstadt, Sengupta, Depp, Palinkas, & Jeste, 2010), two important elements that contribute to life satisfaction (Abu-Bader, Rogers, & Barusch, 2002; Kaushik, 2005). Perceived social support Social support, although studied across all major disciplines, is a concept that carries considerable colloquial meaning. Although it has several definitions, none has 16 been accepted as definitive (Kahn, 1979; Williams, Barclay, & Schmied, 2004). Beginning with the seminal work on social support in the mid-70s, Cobb (1979) defined social support as communicating caring, purely informational, which leads the recipient to "believe that he is cared for and loved, esteemed, and a member of a network of mutual obligations" (pp. 93). This definition, however, seems to emphasize providing emotional assistance to others. In an attempt to offer a holistic meaning of the concept, scholars have extended the definition offered by Cobb to include the provision of material aid. Kahn (1979) considered social support as "interpersonal transactions that involve one or more of the following: expression of a positive affect of one person toward another; the affirmation or endorsement of another person's behavior, perception or expressed views; the giving of symbolic or material aid to another" (p. 85). Similarly, House (1981) defined social support as "personal-level exchanges that involve the expression of affect, the provision of goods and services, and information relevant to one's self-evaluation" (p. 39). Antonucci, Birditt, and Akiyama (2009) emphasized the bidirectional nature of social support and defined social support as the provision or receipt of something (exchange), often including aid, affect, and affirmation, considered to be needed by the provider, recipient, or both. Providing a more simplistic meaning of the concept, Enkenrode and Gore (1981) described social support in terms of number of friendships, proximity to relatives, and involvement with organizations. This definition, however, appears to emphasize structure rather than function (support) of relationship. The above conceptualizations suggest that social support is dynamic and 17 multidimensional. Although the lack of agreement concerning these definitions of social support has produced inconsistences and lack of comparability among studies (Heitzman & Kaplan, 1988; Williams et al., 2004), a closer examination of these definitions reveals two major aspects of social support; the structural (the medium through which support is offered) and the forms or types of support. Three major forms of social support can be identified from the above conceptualizations-affective or emotional, instrumental or practical, and confidant or informational support. Affective support is considered as the most important form of social support, emotional or affective support refers to the expression of love, sympathy, caring, trust, and acceptance of an individual (House, 1981; Wong, Yoo, & Stewart, 2005). Instrumental support includes actions intended to help meet individual's needs, such as providing financial assistance, offering shelter, or services needed to enhance the living condition of an individual (Semmer et al., 2008). Confidant support refers to having a partner with whom secrets are disclosed or private matters discussed (Broadhead et al., 1988; Wong et al., 2005). When looking at social support, it is important to not only consider the type of support but also the amount and the sources of support (Gurung et al., 2003; Thoits, 1982). Variations exist in source, type, and amount of support available, with the latter known to increase in old age (Gurung et al 2003). Support can come from many sources, such as family, friends, neighbors, or even the government (Gurung et al., 2003; Nurullah, 2012). These sources constitute the social support systems (Thoits, 1982). Research has shown that some types of support can only be provided or obtained within certain relationships. It is argued that when the same form of support is obtained or 18 provided by different sources the support may not have the same impact (Gurung et al., 2003; Thoits, 1982). Findings of studies suggest instrumental support is more often provided by family members while emotional support and companionship for the most part are provided by friends (Burke, n.d.; Gurung et al., 2003). Felton and Berry (1992) found that emotional support greatly improved older adult's wellbeing when provided by friends but not when provided by family. However, they also found that confidant support contributed more to the wellbeing of the receiver when provided by family than when provided by friends and neighbors. In the literature, social support is measured either as a perception that a person has assistance available, or an actual occurrence of assistance, often considered as enacted support (Gurung et al., 2003; Lakey & Orehek, 2011; Nurullah, 2012). Due to measurement difficulties, however, the majority of empirical studies have focused more on perceived availability of support rather than actual receipt of support. In many studies, no association was found between provided support and health or receiving support and poor health (Gleason, Iida, Shrout, & Bolger 2008; Lakey & Orehek, 2011; Lakey, Orehek, Hain, & VanVleet 2010; Uchino, 2009). In light of these methodological constraints and empirical limitations, perceived rather than enacted support was examined in this study. Social support is an important construct because of its association to an array of health outcomes (BCMH, 2004; Cohen & Wills, 1985; Cummings & Kropf, 2009; Dimatteo, 2004; Fiori et al., 2006; Lakey & Orehek, 2011; Uchino, 2006; Uchino, 2009). It has consistently been found to be associated with improved health status of older adults. This typically is explained as the result of supportive actions older adults receive 19 from others that moderate the effects of stress associated with aging (Lakey & Orehek, 2011). The perception that family, friends, and neighbors will offer support (perceived support) in times of need is consistently linked to lower levels of distress and loneliness (Chen et al., 2013; Cohen & Wills, 1985), improved cardiovascular biomarker including heart rate, and both systolic and diastolic blood pressure (Thorsteinsson & James, 1999), reduced depressive symptomatology (Schwarzer & Guttierre-Dona, 2005), and reduced mortality among older adults (Shaw et al., 2007). Other studies have also found perceived social support to be associated with treatment and medication adherence among older adults (Cobb, 1979; Dimatteo, 2004; Fiori et al., 2006; Heitzman & Kaplan, 1988). In other studies, however, no evidence was found for the positive impact perceived support is believed to have on the health and wellbeing of older people (Bolger & Amarel, 2007). Since perceptions are often a reflection of lived experience, the results of studies indicating no positive association between perceived availability of support may be a function of one's history of support receipt. It is reported that some supportive behaviors may even be deleterious to the recipient, as they often contribute to feeling of indebtedness and lower self-esteem (Lakey & Scoboria, 2005; Nurullah, 2012). Scholars have attempted to uncover the processes by which perceived social support and health are related. Although some studies have postulated a moderating role of enacted support (Lakey et al., 2010), health behavior (Uchino, 2006), and coping and appraisal (Ben-Zur & Michael, 2007; Frazier, Tix, Klein, & Arikian, 2000; Uchino, 2009) in the association between perceived support and health, results did not support these hypotheses (Ben-Zur & Michael, 2007; Frazier et al., 2000). 20 However, Lakey and Orehek's (2011) work on relational regulation theory, which is premised in the idea that social interaction is the medium through which support is exchanged, is promising. This theory posits that affect, action, and thought of participants in interaction are regulated both by the individual and through relational influences, which occur primarily on a day-to-day basis. Relational regulation occurs through conversation and shared activities that elaborate on recipient's cognitive representation of relationship and quasi relationship. Perceived support is based primarily on relational regulation of affect through day-to-day interaction. Relational regulation theory offers support for the direct effect hypothesis of social support, suggesting that individuals who are actively involved with others will report higher perceived social support and have good health. However, as a relatively new theory, it still needs further examination. Health The quality of a person's life may be considered with reference to its richness, completeness, and contentedness. A range of factors including good physical and mental health, education, financial security, secure occupational environment, spirituality, and strong, supportive social relationships contribute to the overall health of a person (Juniper & Styles, n.d).. Related to health, and often used interchangeably, is the concept of wellbeing (DHHS, 2012; Hanson, 2001). In most studies, health is conceptualized as physical and mental health, and is often indicated with measures such as disease symptoms, disability, functional status, cognitive functioning, functional performance, and participation in physical and social activities (American Thoracic Society, 2007; 21 DHHS, 2012; Golden et al., 2009; La Grow, Neville, Alpass, & Rodgers, 2012; Mann, McCarthy, Wu, & Tomita, 2005; Ware, 2003). These conceptualizations and measures are congruent with the World Health Organization's definition of health, which broadly includes measures of physical, mental, and social wellbeing. Evidence, however, suggests that health in the United States and in other parts of the world is narrowly defined and measured from a deficit perspective, often using measures of morbidity or mortality (Centers for Disease Control and Prevention (CDC), 2011; Hanson 2001; WHO, 1946). To expand its scope to reflect the WHO definition, and for research and policy making purposes, most researchers have now adopted the broad term health-related quality of life (CDC, 2011; Guyatt, Feeny, & Patrick, 1993). Health is a broad, multidimensional concept that refers to the subjective and objective evaluations of physical and mental health, and their correlates such as social relationships and functional status (CDC, 2011; Department of Health and Human Services (DHHS) 2012; Kamphuis et al., 2002; Ware, 2003). A number of personal, economic, social, and environmental factors are known to influence a person's health, although most research has focused on personal (e.g., participation in physical and social activities), and social (social network and social support) factors (Cornwell & Waite, 2009; Perkins et al., 2013; Tay et al., 2013; Uchino, 2013, Uchino, 2006). Over the last few decades, more attention has been focused on health service delivery systems and policies surrounding health care as significant determinants of health (DHHS 2012). Available evidence suggests that health problems become more prominent in late life, affecting quality of life and one's appreciation of life (Abu-Bader et al., 2002; Marjolein et al., 2013; Perkins et al., 2012). Among older adults, health has been 22 examined in relation to social network, social support, sleep problems, as well as chronic and acute conditions (Garcia, Banegas, Graciani Perez-Ragadera, Cabrera, & Rodriguez-Artalejo, 2005; Groessl et al., 2007; Smith et al., 2008). For instance, Garcia and colleagues' (2005) examined the association of social network to health-related quality of life in a population based study of 3600 Spanish non-institutionalized older adults, 60 years and older. Results of the study showed that individuals who were single and lived alone had poor social and mental health status. The results further indicated individuals who reported little or no contact with family members were more likely to obtain worse scores on physical role functioning, body pain, general health, and mental health subscales of the SF-36 questionnaire than those who reported frequent interaction with family. Health scores were also lower among individuals who had little or no contact with friends. Examining the relationship: Social connectedness, perceived social support, and health Research findings on social connectedness and social support in relation to aging and health are mixed. Most findings suggest a decrease in social connectedness following health deterioration in aging and a decline in a person's ability to develop and maintain relationships and social support (Antonucci et al., 2010; Bowling, Edelmann, Leaver, & Hoekel, 1989; Cummings & Henry, 1961; Golden et al., 2009; Kahn 1979; Shaw et al., 2007). Others suggest that aging is marked by a purposeful decrease in social ties allowing for reduction in some types of social relationships or that some forms of support increase with age and others remain relatively stable over time (Adams et al., 2004; 23 Bergeman, Neiderhiser, Pedersen, & Plomin, 2001; Carstensen, 1992; Cornman, Lynch, Goldman, Weinstein, & Lin, 2004; Gurung et al., 2003; Kahn, 1979). Social connectedness, perceived social support, and health are interrelated elements, with each affecting and being affected by the other (see Figure 1). Support exchange is made possible through social ties. Perceptions about social support are usually veridical accounts of specific supportive actions shown through ties with others. It is, however, important to note that not all social relationships involve the exchange of support and that the availability of companionship does not equate provision of support in any form (Antonucci et al., 2009; Ashida & Heaney, 2008; Nurullah, 2012). It is reasonable to assume that large networks and healthy connections with members offer one the opportunity to obtain maximum support. Health is a resource necessary for maintaining social connections (Bowling et al., 1989; Marjolein et al., 2013). Generally, good health in old age ensures the development, maintenance and renewal of social relationships or connections through which support is made available. In the event of significant health problems, development and maintenance of personal relationships are affected in several ways. Disability or illness may decrease older adults' chances of staying active as their mobility becomes affected (Alpass & Neville, 2003; Bowling et al., 1989). Impaired mobility limits one to be physically present around network members. Face-to-face contact therefore reduces and eventually results in loss of relationships. Moreover, decline in mobility prevents people from participating in physical and social activities, two essential elements necessary to maintaining health and developing social relationships (Alpass & Neville, 2003; Marjoleine et al., 2013). Poor mental health has been found to be associated with24 Figure 1: Social relationship and health model 1 The broken lines connecting social support and social connectedness indicates support cannot be obtained without social ties 2 Health represents both physical and mental wellbeing Social Relationship Social connectedness Social support Behavioral mechanisms Health Pathways Social engagement Social influence Access to resources and material goods Health status, both physical and mental Level of need, Ability to reciprocate support, depending on health status Psychobiological e.g., cardiovascular reactivity Health-behavioral e.g., exercise Psychosocial e.g., depression 25 decrease in social contact or interaction as it affects a person's ability to communicate with others (Bowling et al., 1989; Speech Pathology Australia, 2012), and eventually leads to the experience of loneliness (Fees, Martin, & Poon, 1999). Health problems may cause imbalance in the exchange of support. Relationships are interdependent, and all social relationships are formed on the basis of subjective cost-benefit analysis, and critical assessment of alternatives. According to social exchange theory, people tend to keep the support exchanges in their social relationships in equilibrium (Homans, 1958), through the principle of reciprocity (Diekmann, 2004). Health deterioration makes it difficult to give support or reciprocate one received. A relationship marked by an imbalance in support exchange is likely to end (Diekmann, 2004). The case of older adults, however, is quite different as health problems increase their need for and receipt of support (Antonucci et al., 2010; Bergeman et al., 2001; Kahn 1979; Marjolein et al., 2013; Schwarzer & Gutiérrez-Doña, 2005). Older adults are likely to evaluate and perceive as high support if they receive enough resource from others to meet their needs. Social connectedness and perceived social support are known to both directly and indirectly affect physical health and mental wellbeing. The mechanisms by which social relations, social support, and health are related continue to be investigated. Research offers the direct effect and the stress-buffer hypotheses (see Cohen & McKay, 1984; Cohen & Wills, 1985; Gibney & McGovern, 2012), support/efficacy model (see Antonucci et al., 2009), and the relational regulation theory (see Lakey & Orehek, 2011) as providing possible explanations for the association (Cohen & McKay, 1984; Cohen & Wills, 1985; Gibney & McGovern, 2012). By their direct effect, social relationships, 26 working through some behavioral mechanisms such as social engagement, social influence, and access to resources (Berkman, 2007), influence health through psychobiologic (e.g., cardiovascular reactivity, immune system function, blood pressure, stress response), health behavioral (diet, exercise, adherence to medical treatment, smoking, or alcohol use), and psychosocial (depression, self-efficacy, coping, stress management) pathways (Antonucci et al., 2009; Berkman, 2007; Fiori, McIlvane, Brown, & Antonucci, 2006; O'Luanaigh, et al., 2012; Uchino, 2009) (see Figure 1). Larger social networks have been shown to positively impact the health and wellbeing of older adults (Steptoe et al 2013). They have been found to help one prepare for, cope with, and recover from many of distressing life events that characterize old age (Antonucci & Akiyama, 2002). Individuals with limited social networks have been found to be at increased risk of developing cardiovascular disease, infectious illness, mental health problems, and mortality (Antonucci et al., 2010; Cohen, Doyle, Skoner, Rabin, & Gwaltney, 1997; Golden et al., 2009; Holwerda et al., 2012; O'Luanaigh, et al., 2012; Stephens et al., 2011; Tiikkainen, & Heikkinen, 2005). Studies conducted over the last decade offer mixed findings about the relationship between perceived social support and physical and mental health of older adults. Most studies have consistently shown perceived social support to be associated with improved physical and mental health (King, Willoughby, Specht, & Brown, 2006). Perceived support has also been linked to better adjustment to life stress (King et al., 2006), reduced depressive symptomatology (Schwarzer & Guttierre-Doma, 2005), and reduced health morbidity and mortality among older adults (Cummings & Kropf, 2009; Dimatteo, 2004; Fiori et al., 2006; Nurullah, 2012; Shaw et al, 2007). Some studies, however, have 27 reported that some supportive behaviors have no positive effects on health and wellbeing or may even be deleterious to the recipient (Ashida & Heaney, 2008; Nurullah, 2012). Findings indicate that under stressful situations, perceived support is positively related to negative affect and other mental health conditions such as depression and anxiety (Cummings & Kropf, 2009; Lakey & Orehek, 2011). Theoretical framework The convoy model of social relations Kahn and Antonucci's (1980) Convoy Model of Social Relations is one of the general theoretical frameworks underpinning this study. Borrowing from anthropologist David Plath (1975), who used the term ‘convoy' to describe a special closeness that involves supportive interaction, Kahn and Antonucci used the term to denote close social relationships that surround a person, and provide different forms of support essential to the individual's development, health and overall wellbeing. Similar in meaning to convoy in the military, the social convoy protects, defends, socializes, and helps individuals safely navigate the challenges they face through time and space (Antonucci & Wong, 2010; Antonucci et al., 2011). Individuals develop and change over their lifetime. At every point in their life (from infancy to late adulthood), they are members of groups and organizations that help shape their life course (Antonucci & Wong, 2010). The convoy model provides both life span developmental and life course organizational perspectives, for studying the process of aging and other life-course changes in relation to social relationships (Antonucci & Akiyama, 2002; Antonucci & Wong, 2010; Antonucci et al., 2011; Kahn & Antonucci, 1985). Each individual is 28 considered to be going through the life cycle surrounded by a set of people or groups to whom the individual is connected through the exchange of social support (Gurung et al., 2003; Kahn & Antonucci, 1985). A person's convoy at any given time consists of a set of persons or groups on whom the individual depends for support and those who depend on him or her for support. The support received or given may not be symmetrical as they are influenced by factors such as age, health, and social role. The convoy model suggests that people go through life forming social networks which they are motivated to maintain irrespective of age-related changes one might experience and changes occurring in the network composition (Gurung et al., 2003). Individuals evaluate the network from time to time, becoming aware or noting specific strengths and weaknesses network members possess. This knowledge helps them to choose different network members to rely on for different types-emotional, informational, or instrumental-of support or assistance. Effort is made to keep supportive members, while nonsupportive members are avoided (Gurung et al., 2003). The model posits that an individual's convoy is shaped over time by personal (e.g., gender, age, race, and marital status) and situational (e.g., norms, social roles, and expectations) factors, which define the nature of the support relationship one experiences (Antonucci, 2009; Antonucci et al., 2009; Birditt & Antonucci, 2007). These personal and situational factors affect one's health and wellbeing (Antonucci et al., 2009; Perkins et al., 2013). The convoy model identifies three major components of social relations: social networks, social support, and support satisfaction (Antonucci & Akiyama, 2002; Antonucci & Wong, 2010; Antonucci et al., 2009). Together these components help determine the extent to which social relationship is a resource or a risk factor. Social 29 networks, also known as network structure, refer to the objective descriptive characteristics of members in a social relationship such as the size of the network, age and gender of members, frequency of contact, and geographic proximity (Kirke, 2013). Each of these characteristics is an important determinant of health of members in a convoy. Social support refers to the provision or receipt of something, material or immaterial, considered to be needed by one or both parties involved in the support exchange (Antonucci, 2009; Antonucci et al., 2009). Although different forms of support exchanges have been identified (Birditt & Antonucci, 2007; Helgeson, 2003), the convoy model emphasizes three types-aid, affect, and affirmation, all of which are believed to influence health and wellbeing (Antonucci et al., 2009; Kahn & Antonucci, 1980). Individuals are psychological beings and have the ability to evaluate actions. It is important, therefore, to consider their feelings and judgments about support they receive. Act of support is evaluated differently by different people in different situations. In one instance, an act of support may be well received and gratefully appreciated whereas in another instance, it may be seen as unneeded or even demeaning. Recent empirical evidence offers support for many aspects of the convoy model. For instance, findings indicate that both personal (e.g., sex and age) and situational factors (e.g., resource, role expectations, and demands) influence multiple aspects social relations and health (Antonucci & Akiyama, 2002; Gurung et al., 2003; Schwarzer & Gutiérrez-Doña, 2005; Shaw et al., 2007) with clear age and gender differences in network and types of support received. Shaw and colleagues' (2007) examined changes in social relationships throughout late life and found that whereas emotional support remained quite stable with advancing age, informational support increased with age. The 30 results also showed that social contacts with family and friends decreased with age with the higher among men than women. The association between social relations with significant and generalized others and health has been well studied and documented, highlighting the importance of relationships to both mental and physical health (Fiori et al., 2006; García, et al., 2005; Golden et al., 2009; Hawkley, Masi, Berry, & Cacioppo, 2006; Stephens et al., 2011; Williams et al., 2004). Thus, it is important understand the dynamics of social relations and social support as they relate to the aging population. The literature on social support has addressed social relations' direct contributions to health and its ability to moderate the effects of stressful events which may impact one's wellbeing (Antonucci et al., 2009; Cohen, & Wills, 1985; Fiori et al., 2006; Uchino, 2006). This is documented in almost all social and behavioral science literature as the direct-effect and the stress-buffer hypotheses. Direct effect and stress-buffer hypotheses Interpersonal relationships are known to protect people from unhealthy effects of stressful conditions. Lack of positive social relations has been linked to negative psychological conditions such as depression and anxiety (Ashida & Heaney, 2008; Fiori et al., 2006). These negative psychological states, in turn, may influence physical health through behavior patterns or psychological processes that increase the risk for disease (Cohen & Willis, 1985). Social support has widely been used to refer to the mechanisms by which relationship presumptively improve one's health by protecting an individual against 31 stressful events, including stresses often ascribed to the process of aging (Cohen & McKay, 1984; El-Bassel, Guterman, Bargal, & Su, 1998; Gibney & McGovern, 2012; Kahn & Antonucci, 1980). These mechanisms are precisely stated in what have been termed the direct or main-effect and the stress-buffer hypotheses (Cohen & Wills, 1985; Cohen & McKay, 1984; El-Bassel et al., 1998; Gibney & McGovern, 2012). Direct-effect hypothesis The direct-effect, also known as the main-effect hypothesis, suggests that social support has a helpful effect irrespective of whether a person is under stress or not. Stated differently, the hypothesis suggests that social support is advantageous under all conditions, at all times (Cohen & McKay, 1984; El-Bassel et al., 1998). Individuals with stronger social support, according to the direct-effect hypothesis, experience better health and higher levels of wellbeing than people with weak social support (Cohen & Wills, 1985; Gibney & McGovern, 2012). Even though it is well-established and supported empirically, theoretical development to explain the direct-effect hypothesis is lacking (Lakey & Orehek, 2011). Cohen and Wills (1985) suggested the direct-effect hypothesis of social support is evident through an individual's integration in social network that provides one with regular positive experience and stability in one's life situation. The integration provides positive affect and a greater sense of self-worth. Integration may help one to avoid situation with potential consequence of experiencing a psychological or physical disorder. 32 Stress-buffer hypothesis The stress-buffer hypothesis postulates that in the face of stress inducing events the health and wellbeing of individuals with little or no social support is negatively impacted by the stressful events (Cobb, 1976; Cohen & McKay, 1984; Gibney & McGovern, 2012; Kahn & Antonucci, 1980). In other words, the health and wellbeing of those with stronger social support are protected from the deleterious effects of stressful event. Unlike the direct effect hypothesis, the stress-buffer hypothesis appears to be conditional, ‘activated' only when stress is experienced. Thus, social support buffers individual's reaction to a stressful event or enhances one's coping ability (Antonucci et al., 2009). The stress-buffering hypothesis occurs when a person experiences an unwanted and unpredicted life change (perceived as threat) and personal resources are perceived to offer inadequate response to the life change, thereby leading one to seek support from others (Cohen & Wills, 1985; Kahn & Antonucci, 1980). Evidence of its effect is observed when the association between stress and health is weaker for individuals with high levels of social support than for those with low social support. While the literature indicates largely consistent support for the direct effect hypothesis, the stress-buffering hypothesis appears to have empirical limitations, as studies have offered a more nuanced understanding of the hypothesis (Cohen & McKay, 1984; Lakey & Orehek, 2011; Thoits, 1982). Given that the effectiveness and direction of social relations effects may vary depending on the health conditions of a person, social relationships, as well intended as they are, may create or aggravate stressful situations (Antonucci & Wong, 2010; Antonucci et al., 2009; Thoits, 1982). Critics have rejected the proposition of the stress-buffer hypotheses and called for investigation into the 33 theoretical relationship between social support, life events, and psychological wellbeing (Carpenter, 2006; Mezuk, Diez Roux, & Seeman, 2010; Thoits, 1982). Regardless of these shortcomings, the positive effects of direct effect and the stress-buffering hypotheses of social support in relation to health and wellbeing have been well documented (Cohen & Wills, 1985; El-Bassel et al., 1998; Mezuk et al., 2010). Numerous studies indicated that people who receive psychological and material support from family and friends tend to have better health than those with little or no supportive social contact (Carpenter, 2006; Cohen & Wills, 1985; Mezuk, et al., 2010). Social support working through both the direct-effect and stress-buffer mechanisms may affect health outcomes through lessening the "impact of stress appraisal by affecting a solution to a problem, reducing the perceived importance of the problem, soothing the endocrine system so that people are less reactive to perceived stress or by facilitating healthful behavior" (Cohen & Wills, 1985). Social relationships and social support: An integration of theories Social support is an important determinant of health and wellbeing, both for its direct contribution and for its ability to moderate the effects of stress (Kahn & Antonucci, 1980). Drawing from the life course perspective that focuses on the broader context within which people live, the convoy model is proposed as the structure within which social support is given and received (Antonucci & Wong, 2010; Kahn & Antonucci, 1980). The convoy model examines both micro- and macro-level influences that a set of people or groups has on the individual. Such groups may include family, the basic unit of society, school, employment, religious organizations, and the neighborhood (Antonucci 34 & Wong, 2010). The convoy model addresses both the direct and the buffering effects of social support (Antonucci et al., 2009; Kahn & Antonucci, 1985). Social relations, the channel through which support is exchanged, can directly influence physical health and psychological wellbeing at any given time (Antonucci, 2009; Fiori et al., 2006). In addition, when stressful major life changes occur, social relations help moderate the pathological effects through support offered by others and by improving a person's coping skills (Birditt & Antonucci, 2007; Cohen & Wills, 1985; Helgeson, 2003; Uchino, 2006). (See Figure 2.) Research has documented the effects of social relation and social support on psychological or mental health (Carpenter, 2006; Mezuk et al., 2010). In a multi-ethnic study of athereosclerosis, Mazuk and colleagues (2010) evaluated the stress buffering and the direct effect hypotheses of perceived emotional social support on inflammatory markers in a sample of 6814 individuals 45 years and older. The main finding suggested that perceived availability of emotional support had little influence on inflammatory markers, either through direct or stress buffering pathways. Consistent with direct effect hypothesis, low social support was found to be associated with higher levels of C-reactive protein, interleukin, and fibrinogen antigen, which are considered risk factors for cardiovascular morbidity and mortality. Consistent with the stress-buffer hypothesis, the findings showed evidence of high perceived emotional support buffering the association between high stress and C-reactive protein. No other evidence was found for the buffering hypothesis. . 35 Figure 2: Network, support, and health model 1. Network (convoy) is essential for the provision of support 2. Network appears to have a direct relationship with health 3. Effect of support on health is seen through network integration (direct-effect) and in stressful times (stress-buffer) 4. Support seems to have a moderating effect on the relationship between network and health Direct-effect Social support Network Stress-buffer Health Stressful situations Stress appraisal Perceived importance of problem Healthful behavior Necessary at all times Integration Sense of self-worth Self-efficacy 36 Carpenter's (2006) study tested the moderating effect of social support (stress-buffering hypothesis) on the relationship between health status and stress-related psychological outcomes in a sample of gynecologic cancer survivors. The hypothesis that poorer cancer-related health status would be associated with poorer psychological outcomes was clearly supported. While no evidence for moderation was found (not statistically significant), individuals who had strong social support experienced less psychological distress. No direct relationship was found between social support and traumatic stress outcome. The results, however, provided evidence for the stress-buffering hypothesis. Perceived availability of social resources, including support from friends, appeared to be a protective factor against traumatic stress symptoms associated with poor physical health status. The convoy model acknowledges each level of relationship (e.g., family, school) as involving some exchange of support-role demands and responsibilities. In general the model suggests that just as relationship is important and support functional, they can also be dysfunctional. Relationships can provide nurturance and support but they also can expose the individual to physical and psychological threats (Antonucci & Wong, 2010). With the integration of the convoy model, and the direct effect and the stress-buffering hypotheses the negative aspect of relationship and support seem to disappear, suggesting that relationships and support are only beneficial to individual's health and wellbeing. It is important to note that although the support offered to a person may be well intended and serve the needs of the individual, the person may feel pressured to return the support he or she received, a situation that can cause psychological distress for the individual. 37 With respect to the personal and situational characteristics that influence a person's convoy, some studies suggest that characteristics other than social support play direct and moderating roles between life events including stress and health of an individual (Jackson, Knight, & Rafferty, 2010; Yip, Gee, & Takeuchi, 2008). For instance, Yip and colleagues (2008) found that compared to immigrant Asians, ethnic identity moderated the relationship between discrimination and mental health for US-born Asians between the ages of 41-50 years. Similarly, Jackson, Knight, and Rafferty's (2010) study on the stress-buffering role of unhealthy behavior in the relationship between stress and health revealed that for some participants (particularly Blacks), the relationship between stressors and meeting major-depression criteria was weaker among individuals involved in unhealthy behaviors than among those who had not. The authors concluded that by engaging in unhealthy behaviors, which may appear to have protective mental health effects, individuals who live in chronically stressful environments are able to cope better with stressors. What remains unclear is the role personal and situational characteristics played in studies that found support for the moderating role of social support in the association between life events and health. The evidence provided above, however, suggests the need for further investigations to understand the independent contributions of personal and situational factors characterizing one's convoy, and social support in the relationship between life events and health. The convoy model and the social support hypotheses will not be tested; instead, they will be used as conceptual lens describing and interpreting the elements of social relationships-social network or connectedness, and perceived support-and their effects 38 on older adults' physical health and psychological wellbeing. Theoretical and methodological issues in social relationship and health studies Theory, conceptualization, and measurement A substantial body of research offers evidence that concepts used in social relationship studies such as social network, social support, and participation in social activities may serve as a protective mechanism against physical and psychological impacts of life events (thereby improving health) (Cobb, 1979; Cummings & Kropf, 2009; Dimatteo, 2004; Fiori et al., 2006; Lakey & Orehek, 2011; Nurullah, 2012; Thoits, 1982; Williams et al., 2004). However, the evidence must be accepted and interpreted with some level of caution, as there are theoretical and methodological issues with these constructs in the academic literature. Theories are formulated to explain, understand, and predict phenomenon. In most cases, they are formulated to test and advance previous knowledge within the limits of established critical assumptions (Labaree, 2013). While the majority of research on social relationships and health are method-driven, only a few are theory-driven-wherein the researcher applies a particular explicit theoretical framework in order to explore and contextualize the problem they investigate (Public Health Action Support Team, 2011). It has been established that these concepts serving as the components of social relationship directly affect health and wellbeing. A limited number of theories, however, exist to explain the mechanisms by which social network and support are related to health and wellbeing. The direct-effect and the stress-buffer hypotheses have been most cited in the academic literature as offering possible explanations regarding the relationship between 39 social network, social support, and health. As noted earlier, while the direct-effect hypothesis has received empirical validation (Antonucci et al., 2009; Lakey & Orehek, 2011), the majority of studies have found little or no evidence for the stress-buffering hypothesis (Carpenter, 2006; Mezuk et al., 2010). Scholars continue to investigate the mechanisms, and Lakey and Orehek's (2011) recent work on relational regulation theory is considered promising. However, as a relatively new theory, it needs to be thoroughly examined. Methodologically, relationship studies are riddled with conceptual and measurement problems. Conceptual problems include problems with conceptual definitions and boundary specification. Measurement problems include nature of concepts studied and inadequate report on psychometric properties. Due in part to the complexity of social relationship phenomena, there is lack of agreement on definition for almost all concepts used in relationship studies (Kahn, 1979; Lubben & Gironda, 2004; Williams et al., 2004). Williams and colleagues (2004), for instance, identified over two dozens of definitions of social support. As a concept, social support lacks a universal definition accepted by all social researchers (Cobb, 1979; Thoits, 1982; Williams et al., 2004). One problem with the various definitions or conceptualization is a lack of consistency and comparability among studies (Williams et al., 2004). Closely related is the problem of concept operationalization that is necessary for measurement purposes. Heitzman and Kaplan's (1988) review of studies assessing methods for measuring social support identified 23 different operational definitions (e.g., social ties, social network, information given, guidance, social interaction, social integration, etc.) for measuring social support. Despite this, many studies on social 40 support have operationalized it as receipt of emotional, informational, or instrumental support. The problem with these operational terms is the overlap in meaning or understanding of these forms of support, thereby making it difficult to distinctively assess the contribution of each to health and wellbeing of an individual. For instance, the act of supporting one financially, considered a form of instrumental support, may connote an expression of love and thus the provision of emotional support, Level of connectedness is often measured by network size, frequency of interaction with others, and participation in social activities (Cornwell & Waite, 2009; Cornwell, 2008; Shaw et al., 2007). Deciding where one's social network begins and ends, which network size is adequate for the development and wellbeing of the individual, who provides better support to whom and in what situation, and what level of involvement in social activities is healthy for the individual has proven challenging in relationship studies (BCMH, 2004; Dickens, Richards, Greaves, & Campbell, 2011; Tilburg, 2002; Voils et al., 2007). Small network size and less participation in social activities have been used in the literature as indicators of low level of connectedness or integration (Ashida & Heaney, 2008; Cleak & Howe, 2003; Voils et al., 2007). Some research and theories, however, reject this position, claiming that quality is more important than quantity in relationships (Besser & Priel, 2008; Bradley & Cafferty, 2001; Tejeda, 2008; Teo, Choi, & Valenstein, 2013), and regarding that satisfaction is more important than the number of activities one participates in (Blace, 2012; Eakman, Carlson, & Clark, 2010; Levasseur, Desrosiers, & Whiteneck, 2010). Most research on social relationships requires participants to give a general rating of their support network, rather than rating specific support providers. General measures 41 are used, as researchers are unable to distinctively identify provider, recipient, and relational influences. Consequently, the association between a general measure of perceived support and health reflects some unknown combination of social influences and support recipient personal characteristics. Respondents make summary judgments of their social network on rules that seem to equalize supportiveness across different providers. It therefore becomes difficult to ascertain who provides better support to whom and in what situation. Concept measurement in relationship studies presents a challenge for most researchers. Because of their qualitative and quantitative nature, concepts used in relationship studies are sometimes difficult to study. Quantitative measures offer the opportunity to examine a particular construct in a large sample; it is obvious, however, that the rich meaning of the construct may be missed as personal expressions are not a characteristic of quantitative measures. For instance, in trying to assess the strength of one's social ties, it is not enough to inquire of respondents the size of their social network, but also to find out if the size of network matters to them and reasons they offer to support their claims. Similarly, frequency of contact either directly (e.g., face-to-face) or indirectly (e.g., telephone) may serve, and has been used in studies, as an indicator of tie's strength (Voils et al., 2007). It is important to note that dwelling on this quantitative measure, one loses the meaning of what it means to be strongly connected to another. What is important, therefore, and needs maximum attention is the need to assess concepts in relationships studies from both quantitative and qualitative standpoints. Several studies report different instruments or scales used to assess these concepts. While the validity and reliability of most instruments are reported in the 42 literature, several remain unreported (Asante & Lundahl, n.d). In the words of Lubben and Gironda, (2004) most instruments used in relationship studies have "unknown or unreported psychometric properties" (p. 20). Researchers consider the general lack of attention to reporting the validity and reliability analysis of most assessment instruments worrisome. Without reports of instrument validity and reliability, it becomes difficult to ascertain whether or not the instruments used actually measured what they were intended to measure and how reliable the instruments were in providing results that are consistent and trusted. This results in difficulty accepting the findings of studies as true and reflecting the situation in the real world. The development of valid and reliable indicators of the concepts is worth considering. Items such as presence or absence of spouse, friends, or confidants, living arrangement, frequency of contact with other, number of people seen within a certain time frame, and the number of social activities one participates in have largely been used in studies examining social relationships. These measures are used as indicators of social connectedness, the level of integration, and in most instances, measures of support one receives (Ashida & Heaney, 2008; Cornwell & Waite, 2009; Voils et al., 2007). Ideally, each concept would have precise conceptual and operational definitions, with little or no room for overlap. The review of previous studies suggests social relationship is an important element in the life of the older adults. Its impact on the physical health and mental wellbeing continues to be of interest to scholars, hence the significant number of studies done in this area of enquiry. Theories and models have been developed, and hypotheses formulated, as the review suggests, with the idea of furthering the understanding of the 43 association between social relationship and health of an individual. Multiple unexplored or less explored areas in this association need to be studied to add to existing knowledge on social relationships and their association to health and wellbeing. The current study aimed to investigate and understand the individual contributions of social ties and social support to the health of the adult population and to contribute to practice, policy, and knowledge development in this area. CHAPTER 3 RESEARCH METHODS This chapter addresses the quantitative approaches and analytic strategies that were used to study the specified research questions and find support for the stated hypotheses. First, the Utah Fertility, Longevity, and Aging (FLAG) study-the original data source for the current study-is summarized. Next, the current study's design and sample are described, the four research questions and hypotheses guiding this study are restated, the variables and measures from the FLAG study relevant to the current study are reviewed, and preliminary analyses (conducted to ensure no violation of the assumptions of normality, linearity, and homoscedasticity) are presented. Finally, the quantitative analytic strategies used in the study are discussed. Fertility, Longevity, and Aging (FLAG) study Background and purpose The FLAG study, an observational longitudinal study, is composed of a statewide multiple statistical analysis of collected and existing medical and demographic records of geographically stable older adults. The study began in 2004 and data collection is ongoing (FLAG study protocol, n.d.). Evidence available suggests humans differ widely in their age at death and health status over their life course. The FLAG project is 45 premised on the hypothesis that a constellation of factors, both genetic and environmental, influence the rate of aging and longevity and attempts to test this claim by identifying families known to have exceptional longevity on whom to measure epidemiologic, social, cognitive, psychological, and molecular traits believed to be associated with aging and longevity. Sample and data collection FLAG utilizes both primary and secondary data. Primary data include the use of blood samples, clinical exams, and questionnaires to obtain information relevant to the study. Secondary data include information on medical and demographic records of subjects obtained through the Utah Population Databases (UPDB). The first wave of the FLAG project had two main phases. The first phase primarily consisted of a series of statistical analyses conducted by the researchers on existing records in UPDB and from Centers for Medicaid and Medicare Studies (CMS) to identify subjects eligible for the study. The second phase involved recruiting families with excess longevity and an age-sex matched control group (i.e., individuals without characteristics of longevity) based upon statistical analyses completed in phase I. Prior to obtaining informed consent, the mini-mental state examination (MMSE) was administered to assess whether prospective subjects were appropriate candidates for inclusion in the FLAG study. Primary data were collected from the two groups identified above in the second phase (FLAG study protocol, n.d.). From the identified exceptionally long-lived families, 900 participants were recruited and enrolled in the study. FLAG includes 500 exceptionally long-lived (EL) 46 persons (proband group) who are approximately 90 years and older, and 400 of their offspring and nieces/nephews (offspring group) who are estimated to be between 50 to 75 years of age. Two hundred individuals were also identified from the UPDB and serve as the age-sex matched control group for both the proband and the offspring groups. Data were collected on multiple variables from the proband and offspring groups and the matched control group, including but not limited to the following: socio-demographic characteristics, health, medical, and reproductive history, cognitive functioning, depression, social network and support, religion, and an array of clinical measures such as hearing, vision, grip strength, blood pressure, pulse, heart rate, lung functioning, height weight, body temperature, and deoxyribonucleic acid (DNA). Institutional Review Board (IRB) Data were collected with adherence to policies and procedures regarding the protection of human subjects (FLAG study protocol, n.d.). A two-part IRB request regarding informed consent was received. The first part was a waiver of consent for use of existing UPDB data and medical diagnoses data from CMS. The second part was approval granted by the University of Utah Institutional Review Board for obtaining primary data from human subjects. Current study Design This cross-sectional study utilized secondary data from the first wave of data collected in the FLAG study. Data on social connectedness, perceived social support, and 47 health of older adult were analyzed with the purpose of understanding the relationship between the dimensions social connectedness and perceived social support and health. Study sample The study sample was comprised of participants, ages 50 years and older, from the offspring group in the FLAG study. Inclusion criteria included age (50+), and having data on social connectedness and perceived social support, the two predictor variables examined in this study. A total of 325 participants meeting these inclusion criteria were involved in the current study. Research question and hypotheses The current study was undertaken to examine the association between social connectedness, perceived social support, and physical and mental health of older adults. The study further aimed to determine the effect of perceived social support on the association between social connectedness and health of older adults. To investigate these associations, the study addressed following questions and hypotheses using a set of health, social, and demographic variables from the FLAG study:  (Q1) Are there associations between the dimensions of social connectedness, perceived social support, and physical and mental health of older adults? Hypothesis 1: Dimensions of social connectedness (network and satisfaction with network) and perceived social support (affective, confidant, and instrumental support) will be positively associated with physical and mental health of older adults. 48  (Q2) Are there differences in how the dimensions of social connectedness and perceived social support relate with the physical and mental health of older adults? Hypothesis 2: Compared to the dimensions of social connectedness, higher scores on the dimensions of perceived social support will correspond with self-rated good physical and mental health scores.  (Q3) What dimensions of social connectedness and social support are important to physical and mental health of older adults? Hypothesis 3: Compared to the dimensions of social connectedness, the dimensions of perceived social support will be significantly stronger predictors of self-rated physical and mental health.  (Q4) Does perceived social support moderate the relationship between social connectedness and physical and mental health of older adults? Hypothesis 4: Perceived social support will moderate the relationship between social connectedness and physical and mental health of older adults. Variables Data for this study were based on self-reported answers of older Utahns who participated in the FLAG project. Variables addressed included the following: (1) social connectedness; (2) perceived social support; (3) physical and mental health; (4) depression; and (5) socio-demographic characteristics (age, gender, marital history, living arrangement, religious affiliation, religiosity, and socio-economic status). These variables were grouped under predictor, criterion, and covariate variables. 49 Predictor variables Social connectedness A participant's social network, measured in the FLAG study with the Duke Social Support Index (DSSI), was used as the social connectedness measure in the current study. Designed for use with older adults, the DSSI offers a measure of the level or degree of a person's connectedness with others-family, friends, and neighbors (Landerman et al., 1989; Pachana, Smith, Watson, McLaughlin, & Dobson, 2008). The DSSI has 10 items with 5-point Likert scale responses from 0 = None of the time to 4 = All of the time. Participants responded to items such as ""How many times did you talk to some friend, relatives or others on the telephone in the past week (either they called you or you called them)?" and "Do you feel useful to your family and friends (i.e., people who are important to you?)". The 10 items were further grouped into 2 dimensions measuring, frequency of contact with network members (considered network hereafter), and satisfaction with network. Items on both dimensions were recoded into categorical variables with response categories ranging from 1 = Hardly ever to 3 = Most of the time. Network dimension scores ranged from 2 to 9 with higher scores showing more social contacts. The satisfaction with network dimension scores ranged from 9 to 21. Higher scores indicated greater level of satisfaction with social network. Scores for the overall index ranged from 11-30, with higher score indicating more connectedness. Cronbach's alpha coefficients of .578 and .726 were recorded for the network and satisfaction with network dimensions, respectively. The overall index was found to have a reasonable internal reliability with a Cronbach's alpha of 0.74, and a small to moderate interitem correlation recorded in this 50 study. Construct validity was supported in previous research (George et al., 2010; Goodger, Higganbotham, & Mishra, 1999). Perceived social support Perceived social support was measured with the Duke-UNC Functional Social Support (DUNCFSS) Questionnaire, which was developed to provide a brief assessment of functional social support (Broadhead et al., 1998; Sansoni, Marosszeky, Sansoni, & Fleming, 2010). It is designed specifically to measure an individual's perception of the amount and type of personal social support. The DUNCFSS instrument has 10 items with 5-point Likert scale responses from 1 = Much less than I would to 5 = As much as I would like to). Participants responded to items such as "I get love and attention; I get chances to talk to someone I trust about my personal and family problems. The 10 items were further grouped into 3 subscales (dimensions) measuring affective support, confidant support, and instrumental support, with scores ranging from 2-10, 5-20, and 5-20, respectively. Scores for the overall index ranged from 12-50, with higher scores reflecting higher perceived social support. Cronbach's alpha coefficient of .741, .825 and .686 were recorded for affective, confidant, and instrumental support, respectively. The overall index was found to have an excellent internal consistency with a Cronbach's alpha of 0.86, and a moderate to strong interitem correlations found in this study. (See Table 1.) 51 Table 1: Summary statistics for dimensions of social connectedness, perceived social support, and health measures Scale and dimensions Items in scale Cronbach's alpha Range Ma Social connectedness Network 3 .578 2-9 0.262*** Satisfaction with network 7 .726 9-21 0.309*** Overall indexb 10 .740 11-30 0.233*** Social support Affective support 3 .741 2-10 0.506*** Confidant support 4 .825 5-20 0.542*** Instrumental support 3 .686 5-20 0.433*** Overall indexb 10 .867 12-50 0.425*** Health Physical health 10 .754 10-100 0.364*** Mental health 5 .813 24-92 0.490*** Depression 29 .846 0-29 0.183*** Notes: *p<.05; **p<.01; ***p<.001 a Mean interitem correlation b Overall index represents a combined score of all individual subscales/dimensions 52 Criterion variables Health-physical and mental Physical and mental health were measured with the Medical Outcome Study Short-Form 36 (SF-36) in the FLAG study. The SF-36 comprises a generic, coherent, and easy to administer quality-of-life measure designed to examine functioning and wellbeing in older adults. The 36 items are used to compute 8 domains that primarily measure physical and mental health: physical functioning (PF), role limitations - physical (RP), bodily pain (BP), general health (GH), energy (E), social functioning (SF), role limitations - emotional (RE), and mental health (MH) (McHorney et al., 1993). After recoding, each item is scored on a 0-100 range. A higher score indicates more favorable health status (RAND, 2009). For purposes of the current study, the physical and mental health domains were examined. Examples of items in the questionnaire include: "In general would you say your health is____. Response categories ranged from 1 = Excellent to 5 = Poor. "During the past 4 weeks, how much of the time has your physical health or emotional problem interfered with your social activities (like visiting friends, relatives etc.)." Response categories ranged from 1 = All of the time to 5 = None of the time. Cronbach's alpha coefficients of 0.75, and 0.81, with moderate to strong interitem correlations were recorded for physical health and mental health, respectively, indicating both domains of the SF-36 scale have acceptable internal reliability. (See Table 1.) The validity of the SF-36 scale has been tested in relation to socio-demographic and clinical variables, and it has been proven to be a valid measure (Failde & Ramos, 2000; Findler et al., 2001; Gandek et al., 1998). 53 Depression Depression was assessed with the 30-item Geriatric Depression Scale (GDS). The GDS required a participant to respond by answering "yes" or "no" in reference to how he or she felt over the past 30 days, giving an indication of whether or not the participant is depressed. One point was assigned to each answer and the cumulative score was rated on a scoring grid. The grid set a range of 0-9 as "normal", 10-19 as "mildly depressed", and 20-30 as "severely depressed" (Encyclopedia of Mental Disorders, 2013). Examples of items in the scale include the following: "Are you basically satisfied with your life?; Have you dropped many of your activities and interests?; Do you feel that your life is empty?" (See Appendix for scale.) The GDS has an excellent internal consistency with a Cronbach's alpha value of 0.84 and moderate to strong interitem correlations recorded in this study. (See Table 1.) Covariates Covariates included seven items asking participants about their age, gender, marital status, living arrangement, socio-economic status, and religious affiliation and religiosity. Age was a continuous variable ranging from 50 to 81 years. To examine whether or not the levels of connectedness and support change with aging, age was recoded into categorical variable with three response categories: 0 = 50-59, 1 = 60-69, and 3 = 70-81. Gender was a categorical variable with two response categories: 0 = Male, and 1 = Female. Marital status was a categorical variable with five response categories: 1 = Never married, 2 = Married/Living as married, 3 = Separated, 4 = Widowed, and 5 = Divorced. Since a majority of the participants were married, this variable was recoded54 into a dichotomous variable with response categories: 0 = Not married/single and 1 = Married. In regard to living arrangement, participants indicated number of people living in household, including self. The number ranged from 1 to 9, with 1 indicating living alone. Since a majority of the participants fell between 2 and 9, living arrangement was recoded into a dichotomous variable with response categories: 0 = Living alone and 1 = Living with others. Socio-economic status (SES) measured in terms of family's gross income was a continuous variable with response categories ranging from 0 to 100,000 or more. Three groups of SES were identified: 1 = Poor (individuals making 39,999 or less), 2 = Fair, (individuals making 40,000 to 49,999), and 3 = Good (individuals making 50,000 or more, with the majority falling between 50,000 and 69,999). With a majority of the participants falling in the ‘good' category, individuals in the ‘poor' and ‘fair' categories were put together as a group. SES was recoded into categorical variable with two response categories: 0 = Poor to fair, representing individuals with family gross income less than 49,999, and 1 = Good, representing participants with family gross income of 50,000 or more. Religious affiliation was a categorical variable with six response categories: 1 = Latter-day Saints (LDS), 2 = Protestant, 3 = Catholic, 4 = Jewish, 5 = Some other religion, and 6 = No religion. Religiosity was a categorical variable with five response categories: 1 = Deeply religious, 2 = Fairly religious, 3 = Only slightly religious, 4 = Not at all religious, 5 = Against religion, and 6 = Don't know. Since a majority of the participants considered themselves religious, religiosity was recoded as a dichotomous variable with response categories, 0 = Not religious and 1= Religious. (See Appendix for instruments.)55 Data analysis procedure Preliminary analysis Preliminary analyses were conducted to ensure no violation of the assumptions of normality, linearity, and homoscedasticity. These data were screened for outliers and missing data and were not significant to affect the analyses and results. Correlations among variables were examined. Correlations were weak to strong in strength, ranging from, r = .002 to .721. This suggested the unlikely possibility of multicollinearity, in which case correlation coefficients will be higher (r = .9 and above) (Pallant, 2010; Tabachnick & Fidell, 2001). Since the study aimed to investigate the moderation effect of perceived social support on social connectedness and selected health measures, steps were taken to ensure the conditions needed to allow for moderation analyses to be conducted were met. These steps included estimating sample size needed for sufficient power to detect the moderation effect, transforming predictor and moderator variables, and creating an interaction term. The predictor variable, social connectedness, and the moderator, perceived social support, were continuous variables. Both were standardized or centered so that they had a mean of 0 and a standard deviation of 1. To estimate sample size, the G*Power program was used. It was determined that a total sample of 300 was needed to perform the moderation analysis (Sample size calculation: effect size = 0.0625, α = 0.05, power = 0.90, number of group = 2, predictors = 3, Response variables = 1, sample size needed = 300) (Faul, Erdfelder, Buchner, & Lang, 2009). The product term was created by multiplying the centered predictor (social connectedness) and moderator (perceived social support) variables. This was done with the Predictive Analytics Software (PASW). 56 Statistical analysis The data were processed using the Predictive Analytic Software 18 (PASW 18). Descriptive statistics were used to provide basic information-frequency, percentage, mean, and standard deviation-about the study sample. Descriptive statistics were also used to check variables of interest for any violation of the assumptions underlying statistical techniques used to address the research questions (Pallant, 2010). Inferential statistics were later used to analyze the types and degrees of relationship or association among the variables of interest. In addition to maintaining the individual dimensions of the instruments used to measure the constructs under investigation, summed scores were computed to help with the analysis. Reliability analyses were conducted to test instruments' reliability with the study sample. Correlation analyses were used to examine the strength and direction of relationship between the covariates, the predictor, and the criterion variables. Multiple regression analyses were conducted to examine how well the dimensions (indicators) of social connectedness and perceived social support are able to predict physical and mental health when controlling for the effects of covariates. Since the study aimed at investigating the association between social connectedness, perceived support, and health, it was obvious that participants will vary on all these measures. It was expected that some participants would obtain higher health scores than others, and rank higher on the dimensions of social connectedness and perceived social support, suggesting they were more connected and supported. Group difference on these measures (social connectedness, perceived support, and physical and mental health measures) were tested using Chi-square test for independence for 57 categorical variables and t-test for continuous variables. The moderating effect of perceived social support on the relationship between social connectedness and physical and mental health was tested with multiple regression analysis (Baron & Kenny, 1986; Pallant, 2010; Tabachnick & Fidell, 2001; Trochim & Donnelly, 2008). To control the probability of committing Type 1 error, the significance level for these tests was set at alpha value .05. Analysis outputs in Chapter 4 are presented with tables to facilitate understanding of how data were analyzed and conclusions reached. CHAPTER 4 FINDINGS This chapter provides descriptive data for participants for variables examined in the study. The chapter also presents statistical findings for each research question and hypothesis identified in Chapter 1. Descriptive data Socio-demographic characteristics of study participants The mean age of the sample was 64.89 ± 6.98, with a range from 50 to 81 years. More than half (58.2%) of the participants were female. Most (83.4%) were married. The remaining 16.6% were divorced (3.4%), separated (6.5%), or widowed (6.8%). The majority (71.8%) reported good social-economic status. More than two-thirds (89.2%) indicated they lived with others (spouse, children, siblings). Almost all participants belonged to a religious faith with 94.1% identifying with the Church of Latter-day Saints (LDS) faith. (See Table 2.) This is consistent with the religious composition of the population in the state where the study was conducted. 59 Table 2: Socio-demographic characteristics of study participants Categories N % M(SD) Age -- 325 -- 64.89 (6.98) Gender Male Female 136 189 41.8 58.2 -- Marital status Unmarried/single Married 54 271 16.6 83.4 -- Socio-economic status Poor-Fair Good 87 222 28.1 71.8 -- Living arrangement Alone With others 32 292 9.8 89.8 -- Religious affiliation LDS Protestant Catholic Jewish Some other religion No religion 305 0 3 0 6 11 94.1 0 .6 0 1.9 3.4 -- Note: Because of missing data N is not always equal to 325 60 Mean scores of social connectedness, perceived social support, and health measures Table 3 shows the mean scores of both predictor and criterion variables examined in this study. Social connectedness mean scores of 9.91±1.34 and 19.96±1.26 were recorded for the network and satisfaction with network dimensions, respectively. Mean score for the overall index of social connectedness was 29.75 ± 2.62. Scores ranged from 16-33, with high scores indicating more connections and greater satisfaction with network. Based on the mean scores, participants appeared to have strong social connections, and to be highly satisfied with their social connections. The sample's mean score for the overall index of social support was 41.88 ± 6.84, with scores ranging from 16-50. High scores indicated higher perceived social support. Mean scores for the three dimensions were: affective support = 8.72±1.44; confidant support = 16.67±3.33; and instrumental support = 17.80±2.76. Higher scores reflect higher perceived social support; thus, the mean score suggested participants perceived the support they received from others as good. (See Table 3.) The sample's mean score for depression was 4.53 ± 4.20, which suggested low incidence of depression. Scores for depression also showed less variability because most participants (89.2%) were not depressed. This offered statistical and empirical grounds for excluding depression from subsequent analyses. The sample's mean scores on the SF-36 scale were 84.03 ± 15.22, 7 and 3.65 ± 13.66 for physical, and mental health domains, respectively. Higher scores indicated more favorable health on the above mentioned domains. (See Table 3.) 61 Table 3: Mean scores of social connectedness, perceived social support, and health measures N Mean SD Range Social connectedness Network 310 9.91 1.34 2-9 Satisfaction with network 323 19.96 1.26 9-21 Overall index 325 29.75 2.62 11-30 Social support Affective support 325 8.72 1.44 2-10 Confidant support 325 16.67 3.33 5-20 Instrumental support 243 17.80 2.76 5-20 Overall index 325 41.88 6.845 12-50 Health Physical health 324 84.03 15.22 10-100 Mental health 325 73.65 13.66 24-92 Depression 325 4.53 4.20 0-29 Note: Overall index represents a combined score of all individual subscales/dimensions62 Sample demographics according to the level of social connectedness A Chi-square test for independence was conducted to test the bivariate associations between sample demographic characteristics and the level of social connectedness. Using Yates Continuity Correction, social connectedness was significantly associated with religiosity, X2(1, n = 325) = 15.247, p<.01, phi = .217. (See Table 4.) The results suggested individuals who were connected (65.4%) were more likely to be affiliated with religious organization compared to those who were not affiliated with any religious organization (34.6%). The rest of the demographic (age, gender, marital status, socio-economic status, and living arrangement) variables showed no association with social connectedness. Sample demographics according to the level of support Marital status X2(1, n = 325) = 18.230, p<.001, phi = .237, socio-economic status X2(1, n = 325) = 7.736, p<.01, phi = .166, living arrangement X2(1, n = 325) = 15.217, p<.001, phi = .228, and religious affiliation, X2 (1, n = 325) = 13.941, p<.01, phi = .207 were found to be significantly associated with social support. (See Table 5.) The results indicated a statistically significant difference between the proportions of married (69.4%) and unmarried/single individuals (38.9%) who felt supported. There was a statistically significant difference between the proportions of individuals with poor - fair (51.7%) and good (69.4%) socio-economic status in relation to support. The proportion of people living with others (67.8%) who felt supported was statistically significantly different from those who lived alone (31.3%). The proportion of 63 Table 4: X2-test - Distribution of sample demographic characteristics according to level of social connectedness (n=325) Connected (n=213) Not connected (n=112) Category n % n % X2 P Effect size Demographic Age 50-59 60-69 70+ 50 102 61 63.3 64.2 70.1 29 57 26 36.7 35.8 29.9 1.119 .572 -- Gender Male Female 80 133 59.7 69.6 54 58 40.3 30.4 3.014 .083 -- Marital status Single Married 33 180 61.1 66.4 21 91 38.9 33.6 .122 .726 -- SES Poor to fair Good 52 152 59.8 68.5 35 70 40.2 31.5 1.738 .187 -- Living arrangement Alone With others 18 195 56.3 66.8 14 97 43.8 33.2 .991 .320 -- Religious Affiliation LDS Catholic Some other religion No religion 206 2 2 2 67.5 100 33.9 18.2 99 0 4 9 32.5 0 66.7 81.8 15.247 .002 .217 Notes: LDS = Church of Latter-day Saint64 Table 5: X2-test - Sample demographic characteristics and perceived social support (n=325) Supported (n=209) Not supported (n=116) Category n % n % X2 P Effect size Demographic Age 50-59 60-69 70+ 52 100 57 65.8 62.9 65.5 27 59 30 34.2 37.1 34.5 .273 .872 -- Gender Male Female 85 124 63.4 64.9 49 67 36.6 35.1 .025 .874 -- Marital status Single Married 21 188 38.9 69.4 83 33 61.1 30.6 18.230 .001 .237 SES Poor to fair Good 45 154 51.7 69.4 42 68 48.3 30.6 7.736 .005 .166 Living arrangement Alone With others 10 198 31.3 67.8 22 94 68.8 32.2 15.217 .001 .228 Religious affiliation LDS Catholic Some other religion No religion 201 2 0 5 65.9 100 0 45.5 104 0 6 6 34.1 0 100 54.5 13.941 .003 .207 Notes: LDS = Church of Latter Day Saints65 participants with religious affiliations who felt supported (65.2%) was significantly different from those who were not affiliated with any religious organization (34.8%). Married participants who lived with others, those with good socio-economic status, and those affiliated with religious organizations felt more supported than unmarried participants who lived alone, those who reported poor to fair socio-economic status, and those who were not affiliated with any religious organization. (See Table 5.) Differences in dimensions of social connectedness and perceived social support in relation to physical and mental health Social connectedness Using independent samples t-test, the mean scores of the sample on health variables were compared in relation to the dimensions of social connectedness and perceived social support. (See Table 6.) Results showed statistically significant differences in mean scores on the satisfaction with network dimension in relation to physical and mental health. For physical health, participants with higher scores (M = 85.10, SD = 13.462) on the satisfaction with network dimension were significantly different from participants with lower scores (M = 80.99, SD = 19.339) on the dimension, t (323) = -2.117, p = .035. Magnitude of the difference in means score (mean difference = -4.116, 95% CI: -7.940--.292) was small (Eta squared = .014). In terms of mental health, a statistically significant difference was found between participants who scored higher (M = 76.02, SD = 12.143) on the satisfaction with network dimension than those who scored lower (M = 66.72, SD = 15.637); t (323) = -5.533, p = 66 Table 6: Means score differences in dimensions of social connectedness in relation to physical and mental health (t-test) Connectedness Network Satisfaction with network High (n = 209) Low (n = 101) t High (n = 242) Low (n = 81) t M M M M Health Physical health 84.04 82.97 -.568 85.10 80.99 -2.117* Mental health 73.94 72.20 -1.039 76.02 66.72 -5.533*** Notes: *p<.05; **p<.01;*** p<.001 Effect sizes (eta squared) - .01 = small effect; .06 = moderate effect; .10 = large effect Satisfaction with network and physical health = 0.014; Satisfaction with network and mental health = 0.0867 .001. Magnitude of the difference in the mean scores (mean difference = -9.305, 95% CI: -12.613--5.533) was moderate (Eta squared = .08). No significant differences were found in the mean scores on the network dimension in relation to physical and mental health. Generally, older participants who were more satisfied with their network were more likely to have better physical and mental health compared to those who were less satisfied with their network. Perceived social support The independent samples t-test showed statistically significant differences for all the dimensions of social support in relation to physical and mental health. (See Table 7). For physical health, significant differences were found in mean scores for participants who ranked high on the affective support dimension (M = 86.36, SD = 12.89) and those who ranked low (M = 79.74, SD = 18.056); t (324) = -3.817, p = .001; participants who ranked high on the confidant support dimension (M = 85.89, SD = 13.566) and those who ranked low (M = 81.14, SD = 13.566), t (324) = -2.769, p = .006; and participants who ranked high on the instrumental support dimension (M = 86.50, SD = 12.671) and those who ranked low (M = 81.63, SD = 16.631), t (242) = -2.566, p = .011. Magnitude of the differences in the means scores (mean difference) ranged from -4.747 to -6.620, with small effect sizes, (Eta squared = .023 to .043). In terms of mental health, significant differences were found in mean scores for participants with higher scores on the affective support dimension (M = 76.99, SD = 12.073) and those with lower scores (M = 67.47, SD = 14.334); t (325) = -6.342, p = .001; participants with higher scores on the confidant support dimension68 Table 7: Variations in dimensions of perceived social support in relation to physical and mental health (t-test) Support dimensions Affective t Confidant t Instrumental t High (n = 211) Low (n = 114) High (n = 198) Low (n = 127) High (n = 154) Low (n = 89) M M M M M M Health Physical health 86.36 79.74 -3.817*** 85.89 81.14 -2.769** 86.50 81.63 -2.566** Mental health 76.99 67.47 -6.342*** 76.83 68.69 -5.469*** 76.36 68.72 -4.782*** Notes: *p<.05; **p<.01;*** p<.001 Effect sizes (eta squared) - .01 = small effect; .06 = moderate effect; .10 = large effect Affective support and physical health = 0.043; Affective and mental health = 0.110 Confidant support and physical health = 0.023; Confidant and mental health = 0.084 Instrumental support and physical health = 0.026; Instrumental support and mental health = 0.086 69 (M = 76.83, SD = 12.371) and those with lower scores (M = 68.69, SD = 15.188), t (325) = -5.468, p = .001; and participants who ranked high on the instrumental support dimension (M = 76.36, SD = 11.168) and those who ranked low (M = 68.22, SD = 15.188), t (243) = -4.782, p = .001. Magnitude of the differences in the mean scores (mean differences) ranged from 8.139--9.577, with moderate to large affect sizes (Eta squared = .08 to.11). (See Table 7.) In summary, older adults who perceived receiving more affective, confidant and instrumental support were more likely to have better physical and mental health than those who perceived receiving minimal affective, confidant, and instrumental social support. Social connectedness, perceived social support, and health Results of the study suggested that social connectedness is not always accompanied by social support as evidenced by the moderate correlation between social connectedness and perceived social support (r = .461, p<.01) in this population-based sample of older adults. (See Table 8.) Relatedly, a correlation coefficient of determination, R2 = .173 showed both variables shared 17.3 % of their variance, which suggests that social connectedness and social support are separate constructs that are moderately correlated. The sections below examine the study's four hypotheses in relation to their independent association and relative importance to the three health variables under study - physical health, mental health, and general health. 70 Table 8: Correlations among study variables 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 Age - 2 GD -.002 - 3 MS -.032 -.161** - 4 SES -.217*** -.184*** .318*** - 5 LA -.047 -.072 .721*** .239*** - 6 RG -.163** -.118* .145** .043 -.109 - 7 NW .114* .188*** .041 -.042 -.003 -.192*** - 8 SwN .132* .057 .108 .146** .133* -.073 .375*** - 9 AS .053 .088 .230*** .167** .231*** -.106 .233*** .559*** - 10 CS .113* .077 .132* .141** .147** -.084 .298*** .591*** .707*** - 11 IS .003 -.144* .212*** .246*** .169** -.002 .129* .238*** .579*** .518*** - 12 PH -.139* -.133* .129* .238*** .125* .018 .053 .185*** .240*** .167** .174** - 13 MH .215*** -.102 .112* .108 .086 -.077 .159** .417*** .456*** .365*** .362*** .234*** - 14 DP -.102 .119* -.134* -.241*** -.109 .029 -.254*** -.484*** -.377*** -.380*** -.415*** -.397*** -.682*** - Notes: *p<.05, **p<.01, ***p<.001 Correlation between social connectedness and perceived social support, r = .461, p<.001 GD = Gender; MS = Marital status; SES = Socio-economic status; LA = Living arrangement; RG = Religiosity; NW = Network; SwN = Satisfaction with network; AS = Affective support; CS = Confidant support; IS = Instrumental support; PH = Physical health; MH = Mental health; DP = Depression 71 Research questions and hypotheses Question 1/Hypothesis 1 Dimensions of social connectedness (network and satisfaction with network) and perceived social support (affective, confidant, and instrumental support) will be positively associated with physical and mental health of older adults. Table 8 presents results from correlation analyses testing the association between covariates, predictor, and criterion variables examined in this study. For the predictor and criterion variables, significant weak to moderate positive correlations were found between the satisfaction with network dimension of social connectedness, and physical and mental health. The network dimension was significantly associated with mental health, but not with physical health. Coefficients of significant correlations ranged from, r = .159 to .417, ps<.01. The results generally indicated that higher scores on the dimensions of social connectedness scale corresponded with higher scores on physical and mental health domains. Results also showed significant weak to moderate positive correlations between the dimensions of social support (affective, confidant, and instrumental support), and physical health and mental health. Significant correlation coefficients ranged from, r = .167 to .456, p<.01. Higher scores on the dimensions of social support index correlated with higher scores on the physical, mental, and health domain. In support of hypothesis 1, satisfaction with network, affective, confidant, and instrumental support dimensions were found to be positively associated with physical and mental health of older adult. The association between the network dimension was significant with mental health but not with physical 72 health. Question 2/Hypothesis 2 Compared to the dimensions of social connectedness, higher scores on the dimensions of perceived social support will correspond with self-rated good physical and mental health scores. Logistic regression analyses were conducted to test the impact of the dimensions of social connectedness and perceived social support on the likelihood that study participants would report their health status as good. Two models were tested for physical and mental health. Each model contained a set of five predictor variables, including network and satisfaction with network, and affective, confidant, and instrumental support. Predicted probabilities of good physical health Result for model 1 testing physical health was statistically significant (X2(5, n = 231) = 27.165, p<.001), indicating the model was able to distinguish between participants who reported good physical health. The model with all the predictors explained 15.2% (Negelkerke R square = .152) of the variance in physical health. Affective and instrumental support significantly predicted physical health. Affective support was a stronger predictor of reporting good physical health, with an odds ratio of 3.405, which showed that participants with high affective support scores were more than 3 times more likely to report good physical health than those with low affective support (OR = 3.405 (1.558-7.444). The odds of reporting good physical health was 1.976 for instrumental support received, which indicated that participants with high levels of instrumental 73 support were more likely to report good physical health than those with low instrumental support (OR = 1.97, CI = 1.014-3.848, p<.05). (See Table 9.) Predicted probabilities of good mental health Results of model 2 testing mental health were statistically significant (X2 (5, 231) = 29.564, p<.001), with 16.0% (Negelkerke R square = .160) of the variance in mental health explained by the set of predictor variables. The satisfaction dimension of connectedness significantly predicted mental health (p<.05). The odds of reporting good mental health increased by 3.823 for participants who scored higher on the satisfaction dimension (OR = 3.823, CI = 1.735-8.426, p<.05), which indicated participants who were more satisfied with their network were more likely to report good mental health than those who were less satisfied. (See Table 10.) The results of both models highlight some differences with regards to how social connectedness and perceived social support were associated with physical and mental health. While the satisfaction dimension of social connectedness significantly predicted mental health, the affective and instrumental dimensions of perceived social support predicted physical health. Results of the logistic regression suggested social connectedness and perceived social support may affect aspects of health of older adults differently. 74 Table 9: Logistic regression: Predicted probabilities of good physical health Variable B S.E. Wald OR (95% CI) Social connectedness Network -0.438 .325 1.82 0.645(0.342-1.219) Satisfaction w/network -0.598 .391 2.34 0.550(0.256-1.183) Social support Affective(a) 1.225 .399 9.424** 3.405(1.558-7.444) Confidant 0.136 .415 0.107 0.873(0.387-1.970) Instrumental(b) 0.681 .340 4.001* 1.976(1.014-3.848) Notes: *p<.05; **p<.01; ***p<.001 (a) High levels of affective support (b) High levels of instrumental support 75 Table 10: Logistic regression: Predictors of good mental health Scale dimension B S.E. Wald OR (95% CI) Social connectedness Network -0.221 0.308 0.515 0.802(.438-1.466) Satisfaction w/networka 1.341 0.403 11.061*** 3.823(1.735-8.426) Social support Affective 0.696 0.382 3.322 2.006(0.949-4.240) Confidant 0.020 0.384 1.003 1.020(0.480-2.926) Instrumental 0.403 0.342 1.392 1.497(0.766-2.926) Notes: *p<.05; **p<.01; ***p<.001 (a) Higher levels of satisfaction with network 76 Question 3/Hypothesis 3 Compared to the dimensions of social connectedness, the dimensions of perceived social support will be significantly stronger predictors of self-rated physical and mental health. Physical health Table 11 presents results from hierarchical regression analyses examining the effects of social connectedness and social support on self-rated physical health, after controlling for the influence of socio-demographic variables. Model 1 examined the effects of five of the socio-demographic variables on physical health. The model, with all the variables, was significant, F(5, 213) = 3.862, p = .002, and explained 8.3% (R-squared = .083) of the total variance in physical health. SES (B = 6.717, p = .01) significantly predicted physical health (R-square change = .083, p<.05). The remaining demographic variables were not associated with physical health (p>.05). (See Table 11.) Model 2 examined the effect of network and satisfaction with network (the two dimensions of social connectedness) on physical health, after controlling for the effects of socio-demographic variables. The model was significant, F(7, 211) = 3.168, p = .003. Inclusion of the dimensions of social connectedness did not affect the model's performance in predicting physical health, as neither significantly predicted physical health, R-square change = .012, F change (2, 211) = 1.395, p = .250, after controlling for the effects of socio-demographic variables. The dimensions of social support-affective, confidant, and instrumental support-were introduced in model 3. Their inclusion enhanced the model's performance 77 Table 11: Co-efficients and standard errors from regression of physical health scores on covariate