University of Southern California Dissertations and Theses

Age integration in late life: sociodemographic & psychosocial correlates of intergenerational-only, peer-only, and age-integrated social networks

(USC Thesis Other)

Age integration in late life: sociodemographic & psychosocial correlates of intergenerational-only, peer-only, and age-integrated social networks

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

AGE INTEGRATION IN LATE LIFE:
SOCIODEMOGRAPHIC & PSYCHOSOCIAL CORRELATES OF
INTERGENERATIONAL-ONLY, PEER-ONLY,
AND AGE-INTEGRATED SOCIAL NETWORKS

by

Carly Jade Roman

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
GERONTOLOGY

December 2021

ii

DEDICATION
In loving memory of my Great Grandma Bea and my Papa Jerry, my best friends and biggest
fans who inspire and encourage me to achieve my goals

iii
ACKNOWLEDGEMENTS
This dissertation would not have been possible without the support and guidance of my
friends, family, and mentors. I’m so grateful for my age-integrated network for showing me
firsthand the benefits of meaningful social connections with people of all ages.
First, thank you to my advisor, Elizabeth Zelinski for taking on a mentorship role early
on and guiding me through a successful NSF fellowship application. I’m so glad I took that first
grant writing class where we bonded over NY, and you got to know me as I shared personal
stories about my grandparents and GlamourGals. Thank you for all of the independent study
courses and weekly meetings, and for always bringing a balanced perspective to my sometimes
overly optimistic view.
I also want to thank my dissertation committee members, Jennifer Ailshire and
Christopher Beam, for guiding me throughout my PhD. Jennifer, I learned more in your classes
than any others – thank you for challenging me to think critically about aging issues, for making
multilevel statistics fun, for modeling how to teach, and for always being available for quick
questions. Chris, when we were both in Liz’ grant writing class in fall 2016, I would not have
guessed that I would be so lucky to have you as a member of my dissertation committee; it has
been my pleasure and good fortune to see you excel in your career and grow a clinical
geropsychology lab that became my “home away from home” and introduced me to some of my
closest USC graduate school friends. Thank you for all the time you took to work with me on
this dissertation and the intergenerational attitudes scale development project.
I have many USC gerontology faculty and staff members to thank for making my PhD
experience so great. I would like to thank Dr. Kate Wilber for being my “unsung” advisor – I am
so grateful that you have included me on community-based projects that have fueled my passion

iv
and given me the experience and confidence to be the “intergenerational expert” that you have
helped shape me into. Thank you to Dr. Eileen Crimmins and Dr. Teresa Seeman for serving on
my dissertation committee (before the pandemic had other plans) – it meant so much to have
feedback and guidance from two leaders in the field. Thank you to Maria Henke for your
generous support of GlamourGals, the Intergenerational Phone Chain, PhD students, and my
personal career development – you always go above and beyond, and I appreciate it so much. I
also want to thank Linda Broder – whether organizing an event or recruiting older adults for the
phone chain, you always made things easier and became a friend along the way. I also want to
thank my PhD cohort, “The Fantastic Four” – Gerson, Kristi, and Laura, I feel so fortunate that
we completed this PhD journey together; thank you for being such great friends.
Thank you to Dr. Angela Duckworth, my undergraduate mentor, who provided me with a
foundation for my career in positive psychology and aging, and who continues to mentor me and
provide me with the best advice – I’m grateful that someone I look up to and admire is only a
phone call away when I need words of wisdom. I also want to thank Dr. Melanie Katzman for
encouraging me to push myself for a PhD and for brainstorming career and dissertation ideas
with me (even when cornered during Thanksgiving dinner!).
To Rachel Doyle, thank you for creating GlamourGals, without which I would not have
realized my passion for creating opportunities for older adults to stay meaningfully connected to
others through intergenerational connections. Because of you and the amazing organization you
founded, I was inspired to pursue my gerontology PhD, and now I will devote my career to
spreading and demonstrating the impact of intergenerational programs like GlamourGals. I’m
proud to also dedicate this dissertation in memory of Gloria De Leon and Ruth Klein – our
intergenerational friendships have had a lasting impact on me.

v
Most importantly, I want to thank my family for their unwavering faith in me as I moved
across the country to pursue and achieve my goals. First, thank you to my parents for making my
move from NY to CA more of a commute, and for having confidence in your adventurous
middle child (even when I wasn’t feeling so adventurous or confident myself!). Second, thank
you to my siblings, Amanda, Zoe, and Wyndam, for keeping me sane and entertained through
the ups and downs over the past 5 years. Amanda and Wyndam, thank you for making me an
aunt to Cameron. His arrival in early April set up the perfect time crunch pressure needed to
complete my dissertation, and I get so much joy and inspiration seeing Cameron grow up with
you two as parents, my parents as grandparents, and my grandparents as great-grandparents, the
best age-integrated family a guy could ask for! To my friends and extended family, I am so
grateful to each one of you for your friendship and love, and for cheering me on throughout this
journey. Special thank you to Aunt Lisi for paving the way as a geriatric social worker and Alec
for being the first grandkid to move to LA.
Finally, to my favorite rock band, “Grammy, Grandma, and the Papa’s + GG” – thank
you for being my source of inspiration and encouragement, for being my guinea pigs for my
intergenerational programs and positive psychology interventions, and for always being there
when I want to talk (and sometimes even sing and play ukulele!). Although we’re sorely missing
two members of the band, I am so proud to complete this dissertation and receive my
gerontology PhD in memory of GG Bea and Papa Jerry – I would not be where I am or who I am
today without our special intergenerational connections.

vi
TABLE OF CONTENTS
Dedication ii
Acknowledgements iii
List of Tables vii
Abstract viii
Chapter 1: Background 1
Age Integration Theory 1
Social Convoy Theory 3
Empirical Social Network Findings 5
Empirical Evidence of Age Integration 7
Specific Goals of Dissertation 8
Chapter 2: Sociodemographic Predictors Of Social Network Age Integration Status In Late Life 10
Introduction 10
Methods 16
Results 19
Discussion 22
Chapter 3: Psychosocial Outcomes Of Age Integration Status: Do Age-Integrated Social Networks
Benefit Older Adults? 31
Introduction 31
Methods 39
Results 42
Discussion 44
Chapter 4: How And Why Does Age Integration Status Change Over Time? Exploring Personal
Characteristics And Major Life Events 53
Introduction 53
Methods 65
Results 69
Discussion 74
Chapter 5: Summary & Discussion 85
Study 1 86
Study 2 87
Study 3 89
General Discussion 90
Conclusion 95
References 97
Appendices 112
Appendix A: Row Percentages, Age Integration Status and Confounding Variables 112
Appendix B: Weighted Multinomial Logistic Regression, Well-Being and Social Engagement
on Age Integration Status Error! Bookmark not defined.
Appendix C: Relationship Types 114
Appendix D: Social Network Patterns, Intergenerational, Peer, and Missing Age Connections
115

vii
LIST OF TABLES
Table 1-1. Weighted Descriptive Statistics (N = 3,642) 28
Table 1-2. Weighted Bivariate Statistics: Means And Row Percentages (N = 3,642) 29
Table 1-3. Multinomial Logistic Regression Predicting Social Network Age Integration Status (N =
3,642) 30
Table 2-1. Weighted Descriptive Statistics 50
Table 2-2. Weighted Descriptive Statistics Stratified By Age Integration Status 51
Table 2-3. Weighted Ordinary Least Squares Regression: Effect Of Age Integration Status On Well-
Being And Social Engagement 52
Table 3-1. Transition Variables (N = 1,484) 81
Table 3-2. Weighted Descriptive Statistics, Baseline And Follow-Up (N = 1,484) 82
Table 3-3. Weighted Descriptive Statistics By Age Integration Status Transition (N = 1,484) 83
Table 3-4. Logistic Regression Predicting Age Integration Status Transition (N = 1,484) 84

viii
ABSTRACT
As older adults live longer and have more opportunities to remain socially integrated with
others, they can connect with individuals not only in their own generation, but also in younger
generations – that is, older adults have more opportunities for age integration within their social
networks. Prior research has compared those with and without intergenerational ties to
understand age integration within older adults’ social networks; however, the lack of
consideration for presence of same-aged peers suggests that age integration of social networks is
only reflected by having intergenerational connections. We argue that a more accurate
representation of age integration in social networks is having both same-aged peers and
intergenerational connections because it considers the individuals’ social integration with several
age groups, rather than just intergenerational connections. This dissertation fills a gap in the
literature by defining age integration status of older adults’ social networks as age-integrated,
intergenerational-only, and peer-only to identify whether, which, and how older adults benefit
from age integration.
The objective of this dissertation is to further our understanding of age integration in
older adults’ social networks, including how age-integrated older adults vary from those with
peer-only and intergenerational-only networks (Chapter 2), whether they derive unique
psychosocial benefits due to their age-diverse social connections (Chapter 3), and how age
integration status changes over time (Chapter 4). This dissertation consists of three studies which
utilize secondary data from the National Health and Aging Trends Study (NHATS) to investigate
sociodemographic, health, and psychosocial correlates of age integration status in a nationally
representative sample of Medicare beneficiaries in the U.S.

1
CHAPTER 1: BACKGROUND
The number of older adults aged 65 and over in the United States is projected to almost
double from 2016 to 2060 (Vespa et al., 2020). With life expectancies predicted to increase from
almost 80 to almost 86 by 2060 (Vespa et al., 2020), opportunities for older adults to remain
actively engaged in their communities and to be more integrated with others is vital to aging in
place healthfully (Riley & Riley Jr, 2000). However, due to the tripartite segmentation of the life
course – children in school, adults in workforce, and older adults in retirement – our society has
become increasingly age-segregated (Dannefer & Feldman, 2017; Hagestad & Uhlenberg, 2005;
Riley & Riley, 1994; Riley & Riley Jr., 2000; Uhlenberg, 2000). The converse of age segregation
is age integration, which according to Riley & Riley’s (1992) theory, refers to both a reduction in
structural barriers separating individuals by age and an increase in intergenerational interactions
(Riley & Riley Jr., 2000).
Age Integration Theory
The two components of Age Integration Theory (Riley & Riley Jr., 2000) – age-related
structural barriers and intergenerational connections – have bidirectional influences on each
other, such that increasing intergenerational contact may lead to a reduction of age-defined
structural barriers, and simultaneously, the breakdown of age-defined structural barriers may
lead to more interactions across age groups. Age-related structural barriers may include
environmental barriers – for example, older adults living in retirement communities may not
cross paths with individuals in younger generations. They may also include psychosocial
barriers, like ageist attitudes that discourage older adults or younger adults from interacting with
individuals in the opposite age group. Reducing these barriers, whether by creating physical
spaces where older and younger adults socialize with one another or improving attitudes toward

2
other age groups, facilitates intergenerational connections. At the same time, having
intergenerational connections may reinforce positive attitudes toward others who are younger or
older, reducing psychosocial barriers and facilitating further structural changes that allow for
intergenerational connections, like participating in intergenerational programs.
These age-integrated social interactions provide practical benefits of social engagement
to both age groups as older and younger individuals can learn from each other and can provide
and receive emotional and instrumental support from one another. In addition, attitudes between
different groups have been shown to improve with greater contact. Given that attitudes toward
aging influence how individuals themselves age (Levy, 2009; Levy et al., 2009),
intergenerational social connections may foster more positive attitudes toward aging and
therefore, more positive aging outcomes. In addition, intergenerational contact allows older
adults to fulfill a key stage of mid- to late-life development (Erikson, 1959), whereby individuals
express generativity, defined as directing care and concern toward the well-being of younger
generations (McAdams & de St. Aubin, 1992). Empirical reviews of intergenerational programs
have found that older adults benefit from improvements in well-being, self-esteem, and physical
mobility, increased social contact and social connectedness, and fulfillment of generativity
(Canedo-García et al., 2017; Giraudeau & Bailly, 2019). As the growing number of aging adults
are willing and able to contribute their time and energy to society, researchers and policy-makers
have suggested that society must engage older generations’ human capital and generative
abilities to benefit both older adults and the younger individuals whom they help (Dannefer &
Feldman, 2017; Fried et al., 2004; Jenkinson et al., 2013; Parisi et al., 2015; Varma et al., 2015).
The studies in this dissertation are designed to evaluate age integration at the individual
level by characterizing older adults’ social networks as age-integrated if they include same-aged

3
peers and intergenerational connections. Although there is scant empirical research on age
integration within older adults’ social networks (Dannefer & Feldman, 2017; Hagestad &
Uhlenberg, 2005), existing studies have identified intergenerational connections as an indicator
of age-integrated social networks (Dykstra & Fleischmann, 2016; Sun & Schafer, 2019;
Uhlenberg & De Jong Gierveld, 2004). One limitation of these studies is the lack of
consideration for presence of same-age peers in addition to intergenerational connections, which
begs the question: Does age integration simply mean having intergenerational connections, or
does age integration mean having a mix of same-age peers and younger individuals in one’s
social network? The interchangeable use of the terms “age-integrated” and “intergenerational” in
prior studies overshadows the opportunity and need to distinguish between those with both
intergenerational connections and same-aged peers (defined as age-integrated) from those with
only intergenerational connections (defined as intergenerational-only). With this more
comprehensive operationalization of age integration status, sociodemographic and psychosocial
characteristics of those who have age-integrated and intergenerational-only networks can be
compared to each other and to those without any intergenerational social network connections
(defined as peer-only).
Social Convoy Theory
Social Convoy Theory is one of many theories that emphasizes the importance of social
connections in late life (Antonucci et al., 2014; Antonucci & Akiyama, 1987; Berkman & Syme,
1979; Berkman et al., 2000; House et al., 1988). Social convoys consist of social relationships
that vary in importance and emotional closeness depending on life circumstances and personal
characteristics. By characterizing social connections based on how close they are to the
individual, researchers can examine a wide range of social relationships, while incorporating a

4
“life-span and multigenerational perspective” that sheds light on how and why social convoys
change over time (Antonucci et al., 2004, p. 353). These relationships have unique functions that
influence health and well-being, while also being shaped by demographic characteristics, like
age, gender, and race.
The demographic and health correlates of social connectedness have been widely
documented through theoretical and empirical evidence (Antonucci et al., 2014; Antonucci &
Akiyama, 1987; Berkman & Syme, 1979; Berkman et al., 2000; House et al., 1988). Even though
the operationalization of social networks and the mechanisms through which social networks
benefit individuals vary slightly across studies, researchers generally agree that social networks
have positive influences on older adults’ mental and physical well-being. Researchers refer to
social connections using several terms (e.g., social integration, social networks, social ties)
(House, Landis, & Umberson, 1988), but for our purposes we borrow the definition given by
Berkman et al. (2000), which states that social networks are “the web of social relationships that
surround an individual and the characteristics of those ties” (p. 847).
According to Social Convoy Theory, life circumstances influence social network
features, which contribute to well-being and health (Antonucci, 2001) through social support,
social influence, social engagement, and access to resources (Berkman et al., 2000). It is
important to understand how social networks vary among older adults with different personal
and life experiences to elucidate how social convoys are shaped, how they change over time, and
how they influence health and well-being. Although age integration status has not been
thoroughly investigated as a social network feature, Social Convoy Theory suggests those with
age-integrated, intergenerational-only, and peer-only social networks may have distinct
sociodemographic and psychosocial characteristics. By examining social convoy features like

5
age integration status over time, it is possible to understand whether and why group differences
exist.
Empirical Social Network Findings
Although social network characteristics vary widely, the identification and measurement
of social networks is especially important, as they are associated with mental and physical health
outcomes. A foundational study by Berkman & Syme (1979) found that those with greater social
integration, as measured by marital status, familial and non-familial contact, and religious and
organizational memberships, were less likely to die nine years later compared to those with
fewer social ties. House and colleagues (1988) reviewed several studies that corroborated
Berkman & Syme’s (1979) finding that less socially connected individuals were at greater risk of
dying.
A more recent review (Holt-Lunstad, Smith, & Layton, 2010) of 148 studies found that
social relationships were associated with reduced mortality risk, regardless of operationalization.
Social relationship survival benefits were consistent among studies that measured functional
aspects of relationships (e.g., whether one feels emotionally supported, receives financial
support, perceives they have support available if needed), and/or structural aspects of
relationships (e.g., one’s living arrangement, number of social ties, participation in social
activities). Complex measures that consider multiple components of social relationships
(e.g., marital status, number of social ties, and participation in social activities) were found to be
more strongly related to mortality risk than those which only consider one simplified factor (e.g.,
married or not). They also demonstrated that social relationships are as important for survival as
quitting smoking cigarettes.

6
These empirical findings extend to mental health, with more socially integrated
individuals reporting greater mental health and well-being (Antonucci et al., 2013). This
relationship was found to be bi-directional, in that better mental health was linked to greater
social engagement later, while at the same time, greater social engagement led to better mental
health (Schwartz & Litwin, 2017). Social networks have also been characterized by their
compositions, with a wider, more diverse and friend-focused network being associated with
greater morale than those with limited, or restricted, social ties (Litwin, 2001). Also, individuals
with greater social capital across different social network types (i.e., friend, congregant, family,
diverse, and restricted) reported being less lonely, less anxious, and happier (Litwin & Shiovitz-
Ezra, 2011).
The associations between sociodemographic characteristics (e.g., race and gender), and
social network features (e.g., composition and relationship type) sheds light on our
understanding of age integration status as a novel social network feature. Given that researchers
have highlighted the familial and/or non-familial nature of social ties in relation to personal
characteristics (Nguyen et al., 2016), and other studies have documented few intergenerational
ties outside of the family (Drury, Hutchison, & Abrams, 2016; Hagestad & Uhlenberg, 2005), it
is possible that older adults with few family members may be less likely to have
intergenerational ties. Taking these findings in consideration with research on race and
familial/non-familial composition of social networks further contributes to our predictions about
age integration status as an outcome. For example, given that African Americans are more reliant
on family than non-family, as they include extended family members in their social networks
more than White individuals do (Nguyen et al., 2016), it is possible that African Americans
would have more intergenerational social ties as well. When examining the proportion of family

7
members in one’s social network, researchers found that compared to Whites, Blacks had smaller
and more familial social networks with greater contact (Ajrouch, Antonucci, & Janevic, 2001;
Antonucci et al., 2013), again supporting the prediction that Black older adults may have more
intergenerational ties. However, this racial difference became smaller with age, as all older
adults, regardless of race, had a greater density of family members and smaller networks with
whom they had less contact. Similarly, getting older has been linked to a shift toward family-
oriented social networks, which may suggest more availability of intergenerational though these
older adults may have withdrawn from non-familial ties (Li & Zhang, 2015). The authors suggest
this social withdrawal could be both a precursor to and a result of negative physical and
psychological health and policies should encourage diverse social networks with non-family
members to ensure healthy social engagement.
Given the well-documented importance of social connections in old age (Antonucci et al.,
2014; Berkman et al., 2000; Charles & Carstensen, 2010) and the increasing opportunities for
intergenerational relationships in our aging society (Hagestad & Uhlenberg, 2005), this article
provides novel information on the presence and predictors of age-integrated social networks.
With a foundational understanding of the characteristics and prevalence of these relationships,
future research can explore the mental and physical health correlates of intergenerational
relationships.
Empirical Evidence of Age Integration
To the best of our knowledge, none of the studies evaluating older adults’ social networks
include information specifically on their age-integrated nature with consideration of peer social
ties in addition to intergenerational ties. The few studies that identify the presence of age-
integrated social networks suggest that cross-generational ties outside of the family are rare. For

8
example, Hagestad & Uhlenburg (Hagestad & Uhlenberg, 2005) cite a survey in which younger
adults reported non-familial discussion partners over 53 at a rate of 3%, while older adults (60
and older) reported only 6% of their non-familial discussion partners under age 36. While
familial relationships across generations are more common (Drury, Hutchison, & Abrams, 2016;
Hagestad & Uhlenberg, 2005) researchers note that “generational ties in family lines have
become more complex.” There is a need to understand the existence of these relationships on a
larger scale and to determine the demographic predictors of these relationships in order to
identify their associated psychological, sociological, and physical health outcomes in the future.
In order to understand age integration as a feature of social networks, it is first necessary
to define and measure age integration by considering intergenerational connections in the context
of one’s social network, which may or may not also include same-age peers. The current body of
research fills this gap in research by redefining and operationalizing age integration status of
older adults’ social networks as age-integrated, peer-only, or intergenerational-only. These types
of age-integrated social networks may provide unique psychosocial benefits to older adults that
are not experienced by those with peer-only or intergenerational-only networks.
Specific Goals of Dissertation
Increased longevity has allowed older adults to include members from younger
generations in their social networks. Known as age-integrated social networks, prior research has
examined intergenerational ties without considering the presence of same-aged peers. The
current study elucidates sociodemographic factors that correlate with older adults who maintain
peer-only, intergenerational-only, and age-integrated networks.
The objective of this dissertation is to develop our understanding of age integration in
older adults’ social networks by identifying sociodemographic predictors and psychosocial

9
outcomes associated with age integration status. In the few studies that have examined age
integration in older adults’ social networks, age integration is defined simply as having
intergenerational connections. However, social integration with individuals in younger
generations only does not reflect the greater level of age integration afforded to older adults
connected with both same-aged peers and intergenerational ties. One unique contribution of this
work is the operationalization of age integration status, which unlike prior research, distinguishes
between those with only intergenerational connections, and those with intergenerational and
same-aged peer connections, when comparing to those without intergenerational connections
(which we refer to as peer-only). This dissertation explores the following research questions:
1. What are the demographic, sociodemographic, and health characteristics associated
with age integration status?
2. What are the psychosocial benefits of age-integrated social networks?
3. How and why does age integration status change over time?
The goals of this dissertation will be achieved by analyzing age integration status in a
nationally representative sample of Medicare beneficiaries in the United States from the National
Health and Aging Trends Study (NHATS). Chapter 2 elucidates sociodemographic and health
factors that correlate with older adults who maintain peer-only, intergenerational-only, and age-
integrated networks in a cross-sectional investigation of NHATS data collected in 2017. Chapter
3 also uses NHATS data from 2017 in a cross-sectional evaluation of age integration status and
psychosocial correlates (i.e., well-being and social engagement). In Chapter 4, we include data
from the first round of NHATS data collection in 2011 for longitudinal investigations of age
integration status over time, including how and why age integration status changes.

10
CHAPTER 2: SOCIODEMOGRAPHIC PREDICTORS OF SOCIAL NETWORK AGE
INTEGRATION STATUS IN LATE LIFE

Introduction
Relative to opportunities for social engagement with people in their own age group,
individuals tend to have fewer opportunities to interact with those in either younger or older
generations (Riley & Riley, 1994). This may be particularly true for older adults, who experience
shrinking social networks and disengagement from roles that connect them with younger
generations (Uhlenberg, 2000). With increased longevity, age-integrated networks may be an
important feature of adults’ milieus that supports greater quality of life for longer, as they
provide age-diverse social integration in the face of age-based exclusion and disengagement
(Riley & Riley, 1994).
According to Age Integration Theory (Riley & Riley, 1994; Uhlenberg, 2000),
intergenerational interactions and social structures influence each other to determine the level of
age integration in society. A fully age-integrated society that has no age-related barriers to social
participation and high levels of cross-generation contact may be impacted by changes in social
structures that separate individuals by age (e.g., older adults forced to retire); At the same time,
having fewer intergenerational interactions may encourage social structure changes that reinforce
age segregation (e.g., older adults in senior living facilities). Shifts toward societal age
integration, and away from age segregation, have positive implications for older and younger
individuals, like enhancing civility among generations and increasing productive aging. At an
individual level, it has been posited that age integration is reflected by the intergenerational
connections one has, and that these connections are influenced by an individuals’ social

11
structures, including availability of and interest in/attitudes toward intergenerational connections
(Drury et al., 2016; Hagestad & Uhlenberg, 2005).
Although a few studies have examined age-integrated networks in older adults (Dannefer
& Feldman, 2017; Hagestad & Uhlenberg, 2005), these studies assume intergenerational
relationships are the only indicator of an age-integrated network (Dykstra & Fleischmann, 2016;
Sun & Schafer, 2019; Uhlenberg & De Jong Gierveld, 2004). One limitation of these studies is
the disregard of same-aged peers in addition to intergenerational connections. We posit that age
integration means a social network that includes a mix of same-aged peers and younger
individuals. The interchangeable use of the terms “age-integrated” and “intergenerational” does
not distinguish between those with intergenerational connections and same-aged peers (defined
as age-integrated) from those with only intergenerational connections (defined as
intergenerational-only). In this way, those who have age-integrated or intergenerational-only
networks can be compared to those who maintain networks of same-aged peers only (defined as
peer-only).
The current study examines age integration status in a nationally representative sample of
Medicare beneficiaries in the United States. The primary aim of this study is to identify
sociodemographic correlates of age-integrated social networks. Stronger social connections
across the lifespan engender better health and well-being (Antonucci et al., 2014; Antonucci &
Akiyama, 1987; Berkman et al., 2000; Charles & Carstensen, 2010; Seeman et al., 2001). Social
connections have been characterized using multiple terms, including social integration, social
networks, social ties (House, Landis, & Umberson, 1988). For present purposes, we define social
networks as “the web of social relationships that surround an individual and the characteristics of
those ties” (Berkman et al., 2000; p. 847).

12
According to the Social Convoy Model (Antonucci et al., 2014; Antonucci & Akiyama,
1987), relationships are multidimensional and can be defined by their closeness, functionality,
quality, and structure. Relationships can be evaluated both subjectively (i.e., closeness and
quality) and objectively (i.e., structure and functionality). Older adults’ social networks vary in
structural characteristics, like network size and composition (e.g., familial and non-familial
members) and functional characteristics, like supportive roles (e.g., receive financial support,
provide emotional support) (Antonucci et al., 2014; Antonucci & Akiyama, 1987). Given the
many dimensions of social convoys, and the complex ways in which social ties change with life
circumstances, it is important to consider that social interactions may also have negative
components, like conflict or pressure to engage in harmful behaviors (Rook, 1984).
In the current study, we focus on age composition of social networks. Older adults have
reported fewer important individuals in their social networks compared to younger individuals
(Cornwell et al., 2008; English & Carstensen, 2014) due to death and lack of replacement of lost
social network members (Cornwell et al., 2008). Younger generations, thus, may constitute
important populations with whom older adults can socially interact. These intergenerational
social relationships may have a large impact on psychological and physical health outcomes by
providing both generations with opportunities to give and receive emotional and instrumental
support (Berkman et al., 2000).
Age Integration in Social Networks
Age composition of older adults’ social networks has been used to characterize age
integration, particularly inclusion of younger members only. Yet, there is a lack of consensus
across studies about how age integration is defined. A recent analysis of older Europeans’ non-
kin social networks evaluated age integration by inclusion of members who were 10 years older

13
and 10 years younger than participants (Sun & Schafer, 2019). Less than 10% of the sample was
age-integrated, and this group comprised mainly of older adults with denser social networks.
Participants with more children were less likely to be age-integrated.
Some have considered prevalence of cross-age friendships with younger individuals
between ages 18 and 30 as age-integrated (Dykstra & Fleischmann, 2016). When quantified this
way, about 31% of older Europeans reported at least two cross-age friendships. Intergenerational
friendships were most common among older adults who lived with younger family members as
well as those who regularly attended religious services. Like Sun & Schafer (2019), the primary
focus of this study is intergenerational non-kin social networks. It differs insofar that Dykstra
and Fleishman use a greater age gap to operationalize intergenerational friendships.
Others, however, documented the mean number of social network members below age
25, below age 35, and below age 45 for three different age groups of older adult participants (i.e.,
aged 55-64, 65-74, or 75-89 years old) (Uhlenberg & De Jong Gierveld, 2004). They further
distinguished whether younger social network members were kin or non-kin and the median ages
of the youngest kin and youngest non-kin social network members reported by older individuals
(Uhlenberg & De Jong Gierveld, 2004). Not only did older adults report increasingly fewer
younger social network members with increasing age, but also, they reported very few non-kin
younger social network members, suggesting that without kin social network members, Dutch
older adults may not have regular contact with people younger than 45 years old. Being older,
employed, single after divorce, and having more non-kin connections predicted greater odds of
having at least one younger non-kin network member, while being male, more educated, and
having a greater percentage of individuals aged 65 and older in one’s neighborhood predicted
lower odds of having younger non-kin network members.

14
Current Study
The present study operationalizes age integration in social networks as the inclusion of
same-aged peers and intergenerational relationships. Demographic (i.e., age, gender, race),
sociodemographic (i.e., marital status, education), and health (i.e., receipt of self-care help,
number of chronic conditions) characteristics are examined as predictors age integration status.
Research suggests that these characteristics influence social capital and social networks, which
may influence age-integrated networks (Ajrouch et al., 2001; Antonucci et al., 2014; Nieminen et
al., 2007; Sarkisian & Gerstel, 2015).
Increasing age correlates with a shift toward family-oriented social networks (Fiori et al.,
2007; Fiori et al., 2008; Litwin, 2001), possibly providing more opportunities to engage with
younger family members. Social networks may become more family-oriented because of
withdrawal from non-familial ties, engagement with new family ties because of new births (Li &
Zhang, 2015), or because same-aged peers die leaving only younger members.
Women are more likely to replace relationships, making them more likely candidates for
intergenerational social engagement compared to men (Rook & Charles, 2017). Women have
larger and more diverse social networks (Ajrouch et al., 2005; Antonucci & Akiyama, 1987),
partly because of their role as “kin-keepers” of the family (Rook & Charles, 2017).
Research on racial/ethnic disparities in older adults’ social networks provides conflicting
findings and is limited by primarily comparing White and Black groups. Black individuals more
often engage with family networks than White individuals (Ajrouch et al., 2001), which likely
include younger individuals. Black Americans also tend to have smaller and more familial social
networks with greater contact than White Americans (Ajrouch et al., 2001; Antonucci et al.,
2014), which may mean older Black Americans networks are intergenerational-only. Older

15
Hispanic and Asian adults often belong to large family networks (Guo et al., 2015), which
suggests that they, like Black Americans, may be more likely to have age-integrated or
intergenerational-only networks. We note, however, that racial/ethnic differences in network size
are not always observed (Sharifian et al., 2019). Overall, minority status older adults may be
more likely to be age-integrated than White older adults given that younger network members
tend to be kin (Uhlenberg & De Jong Gierveld, 2004).
Under our definition of age-integrated networks, married and partnered older adults tend
to include spouses (and partners) as a close social network member (Antonucci et al., 2019).
Given that partners tend to be peers and not intergenerational connections, partnered individuals
may not list other individuals who are at least 25 years younger than them unless they have
children and, possibly, grandchildren. Conversely, widowed, divorced, and separated people may
be more likely to identify younger individuals as close intimates, as they might rely on their
children and grandchildren for support (Sarkisian & Gerstel, 2015).
Educated individuals may have more socioeconomic resources and opportunities for
economic advancement, which may provide opportunities for age-integrated networks. For
example, those who are more educated have been found to have greater social capital (Nieminen
et al., 2007). Given that greater education has been linked to larger and more diverse social
networks, more educated individuals may be more likely to have both peers and intergenerational
connections in their social networks (Van Broese Groenou & Van Tilburg, 2003).
Finally, people with chronic health conditions may have less age integrated networks, as
they might rely on younger individuals who are physically capable of providing them with
instrumental support. Poor physical health may be a barrier to maintaining peer relationships,

16
leading to older adults developing intergenerational-only networks (Litwin & Shiovitz-Ezra,
2011a; Schafer, 2013)
The present analysis will fill an important gap in the literature by developing a more in-
depth understanding of older adults’ social network age integration and the associated
sociodemographic factors. We hypothesize that older participants, women, minority-raced
participants, non-partnered participants, more educated participants, and less healthy participants
will be more likely to have age-integrated and intergenerational-only networks than peer-only
networks.
Methods
Participants and Procedure
This study draws data from the National Health and Aging Trends Study (NHATS),
which was designed to assess daily functioning among distinct subpopulations of older adults
and to understand how economic and social factors influence health and aging of older
individuals, their families, and society (https://www.nhats.org; Kasper & Freedman, 2014). Data
from the first wave of NHATS were collected in 2011 via in-person interviews from a nationally
representative group of Medicare beneficiaries aged 65 or older, with an oversampling of older
and Black individuals. Participants are re-interviewed annually to document changes in daily
functioning and activities over time. The sample was replenished in 2015 to ensure sufficient
sample sizes. Currently, there are nine waves of data collected in NHATS.
Study data come from the seventh wave of NHATS, conducted in 2016. Participants were
excluded if they had no social network members (N = 222), missing social network data (N =
311), missing age information for all social network members (N = 1,146), and/or missing
demographic information (N = 597). This resulted in a sample of 3,642 individuals, ranging in

17
age from 67 to 101. Logistic regression analyses were conducted to determine whether those
excluded from the sample were a random subset of the baseline sample. Findings suggest that
those who were older and who were more educated were less likely to be excluded from the
analytic sample, while those who were female, Black, non-Hispanic, and never married were
more likely to be excluded from the analytic sample.
Study Measures
Age integration status. Older adults were asked to list and describe up to five
individuals with whom they share important things in their lives, such as “good or bad things that
happen to [them], problems [they] are having, or important concerns [they] may have.”
Individuals listed were considered social network members if their ages were reported. The age
of each participant’s social network member was subtracted from each participant’s age to
calculate the age differences between all participants and each person in their social network. If
any of the participants’ social network members was 25 years or younger, participants were
considered to have an intergenerational social network member. The distinction of 25 years
indicating a generational difference was based on prior sociological research (Carlsson &
Karlsson, 1970). Only eight participants had social network members 25 years or older than
them, so we, thus, considered intergenerational connections as those 25 years or younger. If
participants had social network members within 25 years of their age, they were considered to
have peers in their network.
The age integration status variable was derived using the information about
intergenerational and peer network members. Participants were categorized as having peer-only
networks (coded as 1) if all their network members were within 25 years of their age;
intergenerational-only networks (coded as 2) if all of their network members were 25 years or

18
younger; or age-integrated networks (coded as 3) if their network included peers and
intergenerational members.
Independent variables. Demographic information included older adults’ age (in years),
gender (male = 0, female = 1), and race (White, non-Hispanic = 0, Black, non-Hispanic = 1,
Hispanic = 2, other [American Indian/Asian/Native Hawaiian] = 3). Sociodemographic
information consisted of marital status (married/living with a partner = 0, divorced/separated =
1, widowed = 2, never married = 3) and education (high school or less = 0, more than high
school = 1). Finally, health information included whether participants received self-care help
(e.g., help with bathing, dressing, etc.) in the past year (coded as 1), and number of health
conditions (i.e., heart attack, stroke, etc.) ranging from 0 – 9.
Analysis Plan
All analyses were performed using Stata (Version 16.1). Descriptive statistics (i.e.,
means, standard errors and percentages) were conducted with sample weights to account for
sample selection and non-response biases. In addition to conducting descriptive statistics on the
full sample, analyses were stratified by age integration status (i.e., age-integrated, peer-only, or
intergenerational-only) to examine group differences. Chi-square tests of independence were
calculated for all categorical variables to evaluate the associations between demographic,
sociodemographic, and health characteristics (i.e., gender, race, marital status, education,
whether one received help) and age integration status. Adjusted Wald tests were used to evaluate
the associations between continuous variables (i.e., age and number of health conditions) and age
integration status.
Next, multinomial logistic regression were performed. We first examined whether the
independence of irrelevant alternatives condition was met (i.e., group membership is mutually

19
exclusive). This condition ensures that the outcome groups are distinct, such that participants in
one group could not be replaced if they were removed. For example, if one is in the
intergenerational-only group and that grouping is removed, that person cannot conceptually be
placed into any of the other categories. Therefore, multinomial logistic regression is an
appropriate test.
The baseline model only included the effects of demographic predictors: age, gender, and
race. The second model added sociodemographic characteristics to the model: marital status and
education. The third model added health variables: whether self-care help was received and
number of health conditions. The fourth model tested the effect of the interaction between age
and number of health conditions, as these have been found to be positively correlated (Bektas et
al., 2018). Pseudo-R
2
values were compared across models. For each model, the reference
outcome category is peer-only network.
Results
Descriptive Analyses
Weighted descriptive statistics were generated for all variables (Table 1). Over half of the
sample had peer-only networks, 24.64% had age-integrated social networks, and 20.89% had
intergenerational-only networks. Participants ranged in age from 67 to 101 years of age (M =
76.04, SE = 0.12). There were slightly more females (52.05%) than males and most participants
were White, non-Hispanic. The sample consisted of 63.84% married or partnered participants,
23.88% widowed participants, and 11.23% divorced or separated participants. A majority of
participants (61.84%) had more than a high school education. Only 16.41% of the sample
received help with self-care activities in the past year, and on average, participants had about
2.41 chronic conditions (SE = 0.03).

20
Weighted bivariate descriptive statistics were conducted to determine associations
between age integration status and each independent variable. Table 2 provides descriptive
statistics, stratified by age integration status. Chi-square analyses showed statistically significant
associations between age integration status and gender (χ
2
= 78.08, df = 2, p < 0.001), race (χ
2
=
33.19, df = 6, p < 0.01); and marital status (χ
2
= 991.40, df = 6, p < 0.001). Specifically, of all
men, over half had peer-only networks, almost a quarter had age-integrated networks and 15%
had intergenerational-only networks. Of all females, almost half of the sample had peer-only
networks, while even proportions had age-integrated and intergenerational-only networks.
Peer-only networks were most common across all racial/ethnic groups. Minority groups
belonged to intergenerational-only more often than age-integrated networks, whereas White,
non-Hispanic individuals belonged to age-integrated networks more often than intergenerational
networks.
There was not a significant association between age integration status and education (χ
2
=
1.58, df = 2, p > 0.05) or receiving self-care help in the past year (χ
2
= 6.04, df = 2, p > 0.05).
For married/partnered individuals, most (67%) had peer-only networks, over a quarter
had age-integrated networks, and only 5% had intergenerational-only networks. Although those
who were never married predominantly reported peer-only networks at a similar rate to married
individuals, they had more intergenerational-only than age-integrated networks. For those who
were divorced/separated and widowed, most reported belonging to intergenerational-only
networks followed by peer-only and age-integrated networks.
Adjusted Wald tests revealed that age-integrated individuals were, on average, older than
those with peer-only networks, F(1, 56) = 45.77, p < 0.001, but younger than those with
intergenerational-only networks, F(1, 56) = 22.75, p < 0.001. Those with peer-only networks

21
were found to be younger, on average, than those with intergenerational-only networks, F(1, 56)
= 134.63, p < 0.001. Adjusted Wald tests found no difference in number of health conditions for
those with age-integrated networks and those with peer-only networks, F(1, 56) =0.27, p > 0.05,
or intergenerational-only networks, F(1, 56) = 1.48, p > 0.05. However, those with peer-only
networks had fewer health conditions, on average, compared to those with intergenerational-only
networks, F(1, 56) = 4.90, p < 0.05.
Multinomial Logistic Regression Results
Main effects of demographic, sociodemographic, and health variables characteristics
predicted age integration status whereas interaction effects were nonsignificant. Multinomial
regression results are presented from Model 3 (Table 3). Age was associated with 1.08- and 1.05-
times greater odds of intergenerational-only and age-integrated networks compared to peer-only
networks, respectively. Gender predicted significant differences in network status, such that
women were predicted to have 1.37 times greater odds of having age-integrated networks
compared to peer-only networks. Race was not associated with greater odds of having age-
integrated compared to peer-only networks but did predict significantly greater odds of
intergenerational-only networks compared to peer-only networks. Compared to White, non-
Hispanic participants, Hispanic and other-raced individuals were predicted to have 2.24- and
3.37-times greater odds, respectively, of having intergenerational-only networks compared to
having peer-only networks.
Marital status did not predict age-integrated networks but did predict greater odds of
intergenerational-only compared to peer-only networks. Compared to married/partnered
individuals, those who were never married, widowed, or divorced/separated, were predicted to
have 5.79, 19.22-, and 19.54-times greater odds, respectively, of having intergenerational-only

22
networks compared to peer-only networks. Having more than a high school education was
associated with 1.78 times greater odds of having intergenerational-only compared to peer-only
networks; greater education did not significantly predict differences between age-integrated and
peer-only networks. Self-care help significantly predicted lower odds (RRR = 0.68) of having
intergenerational-only compared to peer-only networks; self-care help did not predict significant
differences between age-integrated and peer-only networks. Number of health conditions did not
predict differences in age-integrated or intergenerational-only networks compared to peer-only
networks.
Discussion
The purpose of the current study was to identify who is age-integrated and who is not in
older adulthood. By defining age integration as including both peers and younger members from
other generations, these results provide a more comprehensive understanding of the differences
among those with peer-only, intergenerational-only, and age-integrated networks. In contrast to
prior age integration research, the current study accounted for age diversity in social networks
using demographic, sociodemographic, and health characteristics, while considering that age
integration in social networks may be broader than simply having intergenerational connections.
Indeed, the variation in sociodemographic differences between older adults with
intergenerational and age-integrated social networks in comparison to those with peer social
networks underscores the value of distinguishing between individuals who were previously
combined into one “age-integrated” group by considering presence of same-aged peers.
In a nationally representative sample of older adults in the US, the most common age
integration status was peer-only. Age-integrated and intergenerational-only networks were
equally balanced. Age and gender were the only two characteristics that were associated with

23
greater odds of an age-integrated network compared to a peer-only network. Indeed, older
individuals tend to report fewer social network members in younger age groups but are more
likely to have younger non-kin social network members (Uhlenberg & de Jong Gierveld, 2004).
Although the current study primarily evaluates kin networks, the similarity between previous and
current findings suggests that age-integrated networks may become crucial as people age. It is
possible that this reflects one aspect of Social Convoy Theory (Antonucci & Akiyama, 1987), as
older adults are more likely to have younger individuals in their networks because of healthcare
needs and older friends and same-aged peers pass away. On the other hand, this may reflect a
conscious decision by older adults to include younger individuals in their social networks,
whether in addition to peers or not, as they seek to maintain emotionally meaningful social
connections in later life.
Although we expected women to be more likely than men to have intergenerational and
age-integrated social networks compared to peer social networks, this hypothesis was only
partially supported. Women were more likely than men to have age-integrated compared to peer-
only networks. However, there was no significant difference in the effect of gender on
intergenerational versus peer-only networks. Consistent with their kin-keeper role, women
appear to maintain social ties from different generations, rather than connecting with only
younger individuals or only same-aged peers (Ajrouch et al., 2005; Antonucci & Akiyama, 1987;
Rook & Charles, 2017).
There was also partial support for hypothesis that racial/ethnic minorities would be more
likely than White, non-Hispanic individuals to have age-integrated and intergenerational social
networks compared to peer social networks. Although Hispanic, American Indian, Asian, and
Native Hawaiian older adults were more likely to have intergenerational social networks

24
compared to peer-only networks, they were not more likely to have age-integrated social
networks. Prior research has found that these racial/ethnic groups have more family members in
their social networks than non-Hispanic, White individuals (Ajrouch et al., 2001; AnnW. W.
Nguyen, 2017; Sharifian et al., 2019). In addition, most younger social connections have been
found to be family members (Uhlenberg & De Jong Gierveld, 2004), providing support for the
current finding that Hispanic, American Indian, Asian, and Native Hawaiian older adults are
more likely to have intergenerational ties. Although it is unclear why this does not translate into
having more age-integrated social networks, it is possible that older Hispanic and Asian
individuals’ shared cultural features, such as collectivism and reverence for older adults,
encourages reliance primarily on younger individuals (Guo et al., 2015). Significant findings
were not observed between Black, non-Hispanic and White, non-Hispanic older adults. Post-hoc
analyses revealed that effects of being Black, non-Hispanic were confounded with marital status.
As predicted, compared to married/partnered individuals, those who were
divorced/separated, widowed, or never married were more likely to have intergenerational-only
rather than peer-only networks. Non-partnered older adults’ social networks likely include family
members who are younger, such as children, grandchildren, nieces, and nephews, instead of
including partners/spouses who tend to be peers. This rationale can also be extended to explain
why marital status was not significantly related to likelihood of having age-integrated compared
to peer social networks. That is, partnered individuals had at least one peer connection in their
partners, while the non-partnered individuals by definition did not have this (likely) same-aged
peer partner, and therefore their social networks mostly include children or other relatives who
are in younger generations only.

25
The hypothesis that more educated individuals would be more likely to have age-
integrated than peer-only networks was not supported. On the other hand, findings suggest that
more educated individuals were more likely to have intergenerational-only compared to peer-
only social networks than those with a high school education or less. Although education has
been associated with more diverse social networks (Van Broese Groenou & Van Tilburg, 2003),
this is only reflected in the limited age diversity afforded by intergenerational connections, rather
than the larger age diversity represented in age-integrated social networks.
Those who were in poorer health, as indicated by receiving self-care help and having a
greater number of health conditions, were not more likely to have age-integrated or
intergenerational social networks compared to peer social networks, contrary to hypotheses.
Although it was expected that older adults with greater health needs might have more diverse
social networks as they depend on younger individuals, or both younger individuals and peers,
who can support them as they navigate health issues, this was not supported by the current study.
Contrary to expectation, those who received help with self-care activities were actually less
likely to have age-integrated compared to peer social networks. It is possible that older adults
with greater health needs may not have been able to socially engage with younger individuals
due to their limited ability to care for themselves. One possible explanation for the contradictory
finding is that older adults who received self-care help were assisted by younger individuals, as
hypothesized, but were considered more peripheral members of the older adults’ social convoy
and not a social network member with whom older adults share important information.
Alternatively, the finding may simply reflect that older adults received self-care help from same-
aged peers with whom they share important information, like spouses or partners.
Limitations and Future Directions

26
Although the current study showed who was most likely to belong to age-integrated
social networks – namely, older adults and women – the study is not without its limitations. First,
social network members with missing ages were excluded from the analysis, because they could
not provide meaningful information on whether they were peers or intergenerational social
network members. Missing ages had implications for the composition of kin and non-kin
connections, too. Post-hoc analyses indicate that age information was missing for most non-kin
social connections, but present for most spouses, relatives, siblings, children, and grandchildren
As a result, the current sample consists of older adults with mostly familial intergenerational
social network members, and therefore results cannot be generalized to samples that include
intergenerational non-kin social network members. With complete age information for all social
network members, future investigations can clarify whether other social network members,
including non-familial ones, are peers or intergenerational connections, and can provide a more
accurate and comprehensive understanding of age integration in older adults’ social networks.
Second, the 25-year cutoff value used in the current study may have excluded individuals
that may be considered intergenerational by other definitions or by an age designation close to,
but less than 25 years. For example, intergenerational relationships could be defined by
relationship type, rather than age differences – children, nieces, nephews, and grandchildren
would be considered intergenerational, even if children are less than 25 years younger than the
participant.
Despite these limitations, the present study provides foundational knowledge on the
sociodemographic predictors of older adults’ social network age integration status. By
considering the presence or absence of peers in addition to the intergenerational connections that
typically define age-integrated social networks, we distinguished those with intergenerational-

27
only connections from those with both intergenerational and peer connections.

28
Table 1-1. Weighted Descriptive Statistics (N = 3,642)
All Participants

N = 3,642
M (SE) or %
Dependent Variable
Age Integration Status
Peer 54.48
Intergenerational 20.89
Age-Integrated 24.64
Demographic Features
Age (67-101) 76.04 (0.12)
Female 52.05
Race/Ethnicity
White, non-Hispanic 83.13
Black, non-Hispanic 6.17
Hispanic 6.93
Other (American Indian/Asian/Native Hawaiian) 3.77
Sociodemographic Features
Marital Status
Married/partnered 63.84
Divorced/separated 11.23
Widowed 23.88
Never married 1.05
More than HS Education 61.84
Health Characteristics
Self-Care Help 16.41
Number of Health Conditions (0-9) 2.41 (0.03)

29
Table 1-2. Weighted Bivariate Statistics: Means and Row Percentages (N = 3,642)

Peer

N = 1,797
M (SE) or %
Intergenerational

N = 916
M (SE) or %
Age-Integrated

N = 929
M (SE) or %
Row Total

N
Demographic Features
Age *** 74.55 (0.14) 79.17 (0.38) 76.69 (0.28) 3,642
Gender ***
Male 61.16 15.38 23.47 1,671
Female 48.33 25.96 25.71 1,971
Race/Ethnicity ***
White, non-Hispanic 55.71 19.28 25.01 2,766
Black, non-Hispanic 52.29 27.90 19.82 580
Hispanic 49.02 28.79 22.19 204
Other 40.99 30.26 28.75 92
Sociodemographic Features
Marital Status ***
Married/partnered 66.58 5.26 28.17 2,039
Divorced/separated 39.13 47.98 12.88 402
Widowed 28.89 49.84 21.28 1,150
Never married 64.92 23.02 12.06 51
Education
HS or less 54.06 21.92 24.01 1,573
More than HS 54.76 20.25 25.02 2,069
Health Characteristics
Self-Care Help Status
No self-care help 54.37 21.54 24.09 2,933
Self-care help 55.03 17.57 27.41 709
Number of Health Conditions * 2.37 (0.04) 2.51 (0.05) 2.41 (0.06) 3,642
Note. Chi-square tests of significance were used for categorical variables; Wald tests of
independence were used for continuous variables; * p < 0.05, ** p < 0.01, *** p < 0.001.

30
Table 1-3. Multinomial Logistic Regression Predicting Social Network Age Integration Status (N
= 3,642)
Intergenerational Age-Integrated
RRR 95% CI p RRR 95% CI p
Age 1.08 [1.06, 1.10] *** 1.05 [1.04, 1.07] ***
Female 0.84 [0.63, 1.11] 1.37 [1.11, 1.70] **
Race/Ethnicity
Black, non-Hispanic 1.16 [0.84, 1.58] 0.95 [0.71, 1.26]
Hispanic 2.24 [1.38, 3.63] ** 1.14 [0.72, 1.80]
Other 3.37 [1.06, 5.31] * 1.65 [0.90, 3.05]
Marital Status
Divorced/separated 19.54 [13.23, 28.87] *** 0.77 [0.51, 1.16]
Widowed 19.22 [13.44, 27.48] *** 1.26 [0.97, 1.63]
Never married 5.79 [2.33, 14.38] *** 0.45 [0.15, 1.39]
More than HS Education 1.78 [1.38, 2.30] *** 1.20 [0.97, 1.48]
Self-Care Help 0.68 [0.50, 0.92] * 1.04 [0.80, 1.36]
Number of Health
Conditions
0.93 [0.85, 1.01] 0.97 [0.90, 1.06]
Note. Reference categories: male; white; married/partnered; high school education or less; no
self-care help. * p < 0.05, ** p < 0.01, *** p < 0.001. RRR = Relative Risk Ratio of having
intergenerational or age-integrated social networks compared to peer social networks.

31
CHAPTER 3: Psychosocial Outcomes of Age Integration Status: Do Age-Integrated Social
Networks Benefit Older Adults?

Introduction
Social connections have long been considered an important feature of healthy aging.
According to Social Convoy Theory, social networks consisting of emotionally close others are
dynamic, changing in structure and function in relation to life events and individual
characteristics, such as age, gender, race, health, and marital status (Antonucci et al., 2019). It is
well-documented that social connections confer mental and physical health benefits by providing
opportunities for older adults to remain mentally and physically active, to fulfill social roles and
responsibilities that bolster self-worth, and to provide and receive support (Berkman et al., 2000;
Charles & Carstensen, 2010; Kawachi & Berkman, 2001). Although researchers have focused on
how social network characteristics are associated with positive psychosocial outcomes such as
well-being and social engagement, there are less-considered features that may uniquely
contribute to these outcomes.
The current study offers an alternative perspective on how social networks benefit older
adults by examining a novel social network feature, called age integration status. By definition,
age-integrated networks include same-aged peers and intergenerational connections,
distinguishing them from intergenerational-only and peer-only networks. Exposure to individuals
not only in one’s own age group, but also in younger generations may provide unique
opportunities for older adults to engage in more diverse social settings and become more socially
integrated (Dannefer & Feldman, 2017; Hagestad & Uhlenberg, 2005). Indeed, high social
integration reflects diverse social connections and diverse social activities (Antonucci et al.,

32
2019; Baker et al., 2005; Litwin, 2001; Nguyen, 2017). Although social integration literature
does not typically focus specifically on age integration, the few studies that have investigated age
integration in social networks (Dykstra & Fleischmann, 2016; Sun & Schafer, 2019; Uhlenberg
& De Jong Gierveld, 2004) are limited by only considering presence of intergenerational social
network members without regard for same-aged peers. It is possible that the psychosocial
benefits of social integration (Antonucci et al., 2019; Baker et al., 2005; Fiori et al., 2007;
Litwin, 2001) are similarly experienced by those who are age-integrated, as they too have varied
social ties and experiences via their intergenerational connections and same-aged peers.
We posit that age integration status is related to social engagement and well-being, with
age-integrated older adults having greater well-being and participating in more social activities
than older adults with peer-only or intergenerational-only networks. In order to understand the
hypothesized associations of well-being and social engagement with age integration status, it is
important to first review how social network features, well-being, and social activities are
defined, and how they have been linked to positive late-life outcomes, including better physical
health and reduced mortality.
Social Networks, Well-Being, and Social Engagement
Social networks refer to the social relationships that individuals have, including objective
features, like the number of social ties, and subjective features, like the quality of social
connections (Berkman et al., 2000; House et al., 1988). Life events (e.g., getting divorced,
becoming a parent) and individual characteristics (e.g., age and gender) shape social convoy
features –including whether social network members are intergenerational connections or same-
aged peers – in ways that influence well-being and social engagement (Antonucci et al., 2019;
Antonucci & Akiyama, 1987; Litwin, 2001). Same-aged peer relationships provide opportunities

33
for companionship, sociability, and social engagement to reinforce meaningful social roles
(Berkman et al., 2000; Luo et al., 2020); while intergenerational social ties allow older adults to
fulfill different types of social roles, in which they express generativity, or care and concern
toward younger generations, and experience a sense of purpose and social belonging (Berkman
et al., 2000; Gierveld & Hagestad, 2006; Luo et al., 2020; Uhlenberg & De Jong Gierveld, 2004).
The mechanisms through which intergenerational connections and same-aged peers benefit older
adults may be distinct, as previously described, or shared, as both types of relationships allow
older adults to receive and provide support, and to have social roles that boost self-esteem
(Berkman et al., 2000). Just as those with high social integration (Antonucci et al., 2019; Baker
et al., 2005) and diverse social networks (Fiori et al., 2007; Litwin, 2001) have greater well-
being and social engagement than those who are less socially integrated and more limited in
types of social ties, it is likely that those experiencing age integration in their social network may
reap similar benefits, as the different types of social ties provide more opportunities for and
potential pathways through which social connections can positively influence older adults’ well-
being and social engagement (Berkman et al., 2000).
The primary pathways through which social networks influence health outcomes include
social support, social influence, social engagement and attachment, and access to resources and
material goods (Berkman et al., 2000). For example, social ties may influence older adults to
maintain healthy lifestyles by encouraging exercise and not smoking, which in turn, positively
influences physical health and longevity (Berkman et al., 2000; Kawachi & Berkman, 2001). A
conceptual model presented by Berkman and colleagues (Berkman et al., 2000) highlights the
ways in which upstream factors, like social-structural conditions (e.g., culture, socioeconomic
status) shape social network structures (e.g., size and frequency), which influence downstream

34
factors, like psychosocial mechanisms (e.g., social engagement and social influence) that impact
health through health behavioral (e.g., smoking and exercising), psychological (e.g., self-efficacy
and well-being), and physiological (e.g., transmission of disease and stress responses) pathways.
These factors interact with each other in complex ways (e.g., exercising may support health and
well-being that makes it possible for older adults to socially engage with others, while social
engagement may provide physical and mental stimulation that allows older adults to exercise).
Empirically, social connections have been associated with positive health outcomes,
including reduced mortality (Berkman et al., 2000). Berkman & Syme (1979) found that those
with greater social integration, as measured by marital status, familial and non-familial contact,
and religious and organizational memberships, had lower mortality rates nearly a decade later
compared to those with fewer social ties. Several reviews have confirmed these findings (Holt-
Lunstad et al., 2010; House et al., 1988), while highlighting the range of social network measures
associated with reduced mortality risk, including simple, binary indicators of social contact (e.g.,
married or not, living alone or not) and a combination of social network features (e.g., quality
and quantity of social ties).
These findings extend to mental health, with more socially integrated individuals
reporting better mental health and greater social engagement (Antonucci et al., 2014; Kawachi &
Berkman, 2001; Pinquart & Sörensen, 2000). Richer social networks, as measured by number of
social network members, number of social network members with frequent contact with very
close emotional ties, and number of relationships within different categories (i.e., friend, child,
spouse), had better mental health than less socially connected older adults (Schwartz & Litwin,
2019). The authors suggested that poorer mental health impacts social ties through reinforcement
of negative social interactions that make individuals withdraw from social networks; this process

35
is a vicious cycle in which lack of social connectedness further diminishes mental health. On the
other hand, stronger social connections reinforce feelings of connectedness that encourage
positive psychological changes and further engagement with social ties that bolster well-being.
It is important to consider that cross-sectional studies are unable to clarify the direction of
causality of these relationships. Although social networks may lead to better mental health and
greater social engagement, it is also possible that better mental health and participation in more
social activities determines social network features. It is likely that both statements are true to
some extent: Schwartz and Litwin (2019) showed that greater social connectedness over time
was associated with better mental health over time, and at the same time, greater mental health
over time was associated with greater social connectedness. They hypothesized that social
networks would have a greater effect on later mental health; however, they found that mental
health changes were more strongly associated with social networks, such that individuals who
had greater well-being were encouraged to maintain and strengthen social ties. A longitudinal
examination found that larger social networks were associated with greater engagement in social
activities over a 6-year period (Huxhold et al., 2013).. The increased social stimulation and
emotional support over time mediated the indirect relationship between larger social networks
and improved well-being over time (Huxhold et al., 2013).
Several theories, including Activity Theory (Havighurst, 1961) and Successful Aging
(Rowe & Kahn, 1997), emphasize that continued engagement in interpersonal and productive
activities is necessary for older adults to thrive, mentally and physically. Although older adults
may withdraw from social connections and activities for a variety of reasons, including health
issues, loss of friends and family, retirement, or a conscious decision to disengage, these theories
posit that participation in social activities is a way for older adults to maintain well-being though

36
social interactions and reinforcement of self-worth (Havighurst, 1961). Participating in a wider
range of activities is associated with higher well-being, as it allows for increased purpose
through diverse roles in different social settings (Lee et al., 2018). Empirically, this is supported
by findings that older adults who were active more frequently, as measured by engagement in
productive (i.e., volunteering/helping) and social (i.e., educational/athletic/religious/political)
activities, had greater likelihood of having high well-being, as measured by life satisfaction,
quality of life, self-rated health, psychological distress, chronic diseases, and body mass index
(Vozikaki et al., 2017).
Well-being, defined as optimal psychological functioning and experience (Deci & Ryan,
2008), has also been linked to a wide range of health benefits (Trudel-Fitzgerald et al., 2019).
Those with higher psychological and social well-being were shown to have a slower decline in
physical functions, including walking speed, chair stands, and one-leg balance (Saadeh et al.,
2020), while controlling for confounders by excluding those who had low physical functioning at
baseline. The authors suggest that those with higher psychological well-being perceive their
aging experience more positively, which encourages them to engage in preventive health
behaviors, like eating healthy, exercising, and not smoking. Longitudinal studies have found that
greater well-being is associated with lower cardiometabolic risk scores and lower risk of
coronary heart disease 8 to 11 years later (Boehm et al., 2016), a relationship mediated by health
behaviors, such that people who were happier had better health behaviors, which translated into
reduced risk of heart disease and better cardiometabolic risk levels. Similarly, those with stable
high well-being over time reported better self-rated health, fewer health conditions, and less
disability about 10 years later than those with stable low or stable moderate well-being over time,
while controlling for earlier health characteristics (Ryff et al., 2015).

37
It is possible that age-integrated networks afford older adults unique well-being and
social engagement benefits because the varied interactions tap into different mechanisms through
which social connections benefit older adults. For example, intergenerational connections allow
older adults to express generativity, or care and concern toward others (McAdams & de St.
Aubin, 1992), which has been linked to well-being and reduced mortality over time (Gruenewald
et al., 2007, 2012). In addition, there are practical benefits of intergenerational connections, like
learning about technology that can broaden older adults’ abilities to participate in activities and
connect with others (Uhlenberg & De Jong Gierveld, 2004). Connections with same-aged peers
allow older adults to bond over shared experiences, encourage positive health-related behaviors
(e.g., exercising together, eating healthy foods, not smoking), and relate to one another’s shared
experiences, reinforcing social roles that make older adults feel like they are valued and belong
(Berkman et al., 2000).
Age Integration and Psychosocial Outcomes
The few studies that have examined the age composition of older adults’ social networks
in relation to psychosocial outcomes vary in how they operationalize age integration. In a sample
of European older adults, age integration was defined as upward if participants’ social networks
included at least one non-kin social network member 10 years older than them, or downward if
participants’ social networks included at least one non-kin social network member 10 years
younger than them (Sun & Schafer, 2019). Although only 10% of older Europeans had age-
integrated non-kin social networks in both directions, they found that older adults who
participated in more formal social activities (i.e., clubs, volunteering, courses, community
organizations) were more likely to be age-integrated.

38
Another analysis operationalized older adults’ networks as age-integrated if they had two
or more friendships with non-kin younger than 30 years old (Dykstra & Fleischmann, 2016).
When measured this way, age integration with younger individuals was more common among
older adults who lived with younger adults, attended religious services monthly, volunteered, and
worked with younger individuals (Dykstra & Fleischmann, 2016).
A final study used multivariate analyses to determine the sociodemographic predictors of
having at least one non-kin member at least five years younger, and of having larger age gaps
from one’s younger non-kin member (Uhlenberg & De Jong Gierveld, 2004). They found that
being older, employed, single after divorce, and having more non-kin connections predicted
greater odds of having at least one younger non-kin network member, while being male, more
educated, and having a greater percentage of individuals aged 65 and older in one’s
neighborhood predicted lower odds of having younger non-kin network members. In addition,
those with more non-kin members, who volunteered, went to church frequently, and were
employed had larger age gaps with their youngest non-kin social network member, suggesting
that age integration is more common among those who are more socially engaged.
Although these studies made valuable contributions to our understanding of older adults’
social networks, they also underscore the lack of consensus within and across studies on how to
operationalize the construct of age integration and the lack of consideration for same-age peers
as a feature of age-integrated social networks. The current study uses a more comprehensive
measure of age integration status to test the hypotheses that those with age-integrated networks
have greater well-being and social engagement than those with peer-only and intergenerational-
only networks. Sociodemographic features, including age, gender, race, education, and health,
were also considered in the current study, as they have been linked to well-being (Howell et al.,

39
2007; Pinquart, 2001; Pinquart & Sörensen, 2000; Tang et al., 2019), social engagement (Adams
et al., 2011; Pristavec, 2018), and social network features (Ajrouch et al., 2001; Antonucci et al.,
2014; Fiori et al., 2007; Huxhold et al., 2013; Litwin, 2001).
Methods
Participants and Procedure
The National Health and Aging Trends Study (NHATS) was conducted to assess socio-
economic factors associated with health and aging in a nationally representative sample of older
adult Medicare beneficiaries (https://www.nhats.org). NHATS began in 2011 with data
collection via in-person interviews with a nationally representative group of Medicare
beneficiaries aged 65 or older. To correct for non-response biases, older and Black individuals
were oversampled. Annual interviews are used to track changes in economic and social features
over time. To maintain sufficient sample size, new participants were added in the fifth round of
data collection.
The current study utilizes data from the seventh wave of NHATS, conducted in 2016.
Participants were excluded if they had no social network (N = 222), no social network data (N =
311), missing age information for all social network members (N = 1,146), missing data on
covariates (N = 595), and or/missing data on independent variables (N = 437). This resulted in a
sample of 3,564 participants, ranging in age from 67 to 101. Logistic regression analyses were
conducted to determine whether those excluded from the analytic sample differed meaningfully
from those included in the analytic sample. Participants who had higher well-being, more social
activities, and who received help with self-care activities were less likely to be excluded from the
sample. Conversely, women and Black, non-Hispanic participants were more likely to be
excluded from the sample than men and White non-Hispanic participants, respectively. A priori

40
power analyses conducted using G*Power determined the sufficient sample sizes to detect
significant effects with power of .95, alpha of .05, and seven predictor variables in the well-being
and social engagement ordinary least squares regression models were 141 and 120 participants,
respectively.
Study Measures
Independent variable. Age integration status was derived using information about
participants’ social networks. Specifically, participants were asked to name up to five individuals
with whom they share important things in their lives (i.e., good or bad things that happen to
[them], problems [they] are having, or important concerns [they] may have) to indicate social
network membership. Social network members were excluded from the participants’ social
network if their age was not reported by the participant. Based on prior research suggesting a
generation spans about 25 years (Carlsson & Karlsson, 1970), social network members were
designated peers if they were within 25 years of the participants’ age, and intergenerational if
they were at least 25 years younger than participants. Age integration status was considered
peer-only if participants had only peers, intergenerational-only if they had only intergenerational
connections, or age-integrated if they had both peers and intergenerational connections in their
social networks.
Dependent variables. Well-being was calculated by summing the scores of 8 items
measuring two dimensions of well-being, with items reverse scored such that higher scores
indicated greater psychological well-being. The first subscale consisted of four items measuring
hedonic, or emotional, well-being were endorsed on the following 5-point Likert scale: 1 =
Never, 2 = Rarely (Once a week or less), 3 = Some days (2-4 days a week), 4 = Most days (5-6
days a week), 5 = Every day (7 days a week). Participants were asked how often, during the last

41
month they felt cheerful, bored, full of life, and upset. The second subscale measured
eudaimonic, or psychological, well-being was assessed with 4 items, endorsed on a three-point
Likert scale: 1 = Agree not at all, 2 = Agree a little, 3 = Agree a lot. Participants were asked how
much they agree with statements including, “My life has meaning and purpose,” “I feel confident
and good about myself,” “I gave up trying to improve my life a long time ago,” and “I like my
living situation very much.” Social engagement was measured by counting the number of social
activities older adults participated in during the previous month. Specifically, older adults were
asked whether they participated in the following activities: 1) visit in person with friends or
family; 2) attend religious services; 3) participate in clubs, classes, or other organized activities;
4) go out for enjoyment (e.g., going out to dinner, a movie, to gamble, or to hear music or see a
play); 5) do volunteer work. Dummy codes for participation (coded as 1) or non-participation
(coded as 0) in each activity were summed to indicate total number of social activities, with more
social activities indicating greater social engagement.
Covariates. Demographic information included older adults’ age (in years), gender
(male = 0, female = 1), and race (White, non-Hispanic = 0, Black, non-Hispanic = 1, Hispanic =
2, other [American Indian/Asian/Native Hawaiian] = 3). Sociodemographic information
consisted of marital status (married/living with a partner = 0, divorced/separated = 1, widowed
= 2, never married = 3) and education (high school or less = 0, more than high school = 1).
Finally, health information included whether participants received self-care help (coded as 1)
(e.g., help with bathing, dressing, etc.) in the past year, and number of health conditions, a count
of chronic conditions participants reported (i.e., heart attack, stroke, etc.) ranging from 0 – 9.
Analysis Plan
All analyses were conducted using sample weights to account for non-response biases

42
and to make the sample representative of the Medicare population. First, descriptive statistics
were conducted on the full sample. Then, bivariate descriptive statistics were stratified by age
integration status to determine group differences. For categorical variables, chi-square analyses
were conducted; for continuous variables, adjusted Wald tests were conducted. Next, we
conducted separate ordinary least squares regression analyses for well-being and social
engagement to determine their associations with age integration status and the covariates.
Supplementary analyses (Appendix A) demonstrated the confounding relationships
between age integration status and marital status, parental status, and living arrangement. For
example, widowed individuals rarely had peer-only networks because spouses who in most cases
were same-aged peers, are no longer in their social networks. On the other hand, married or
partnered individuals rarely had intergenerational-only networks, because spouses/partners in
most cases were same-aged peers. Therefore, these covariates were not included in the models.
Results
Descriptive Analyses
All analyses were performed using Stata (Version 16.1). Weighted descriptive statistics
(Table 1) were conducted on the full sample (N = 3,564). The average age of participants was
almost 76 years old. The sample consisted of slightly more women (52.00%), White, non-
Hispanic participants (83.38%), and participants with more than a high school education
(62.10%). Most of the sample had peer-only networks (54.76%), almost a quarter had age-
integrated networks, and 20.72% had intergenerational-only networks. Well-being ranged from
10 to 32, with an average of 26.97 for all participants. Social engagement ranged from 0 to 5,
with participants’ averaging 3.02 social activities.

43
Weighted descriptive statistics were stratified by age integration status to examine
sociodemographic differences among groups. Chi-square analyses demonstrated significant
associations of gender (χ
2
= 77.59, df = 2, p < 0.001) and race (χ
2
= 30.78, df = 6, p < 0.05) with
age integration status. There were no significant associations of age integration status with
education (χ
2
= 1.19, df = 2, p = 0.534) or self-care help (χ
2
= 4.04, df = 2, p = 0.145). Adjusted
Wald tests revealed significant age differences among groups, F(2, 55) = 88.39, p < 0.001, with
peer-only participants being the youngest (M = 74.50), age-integrated participants being slightly
older (M = 76.60), and intergenerational-only participants being the oldest (M = 74.50). Age
integration status was not significantly associated with number of health conditions, F(2, 55) =
2.86, p < 0.001.
Bivariate associations were also examined for outcome variables. Adjusted Wald tests
revealed significant well-being differences among groups for social engagement only, F(2, 55) =
7.27, p < 0.01, with age-integrated participants having the greatest number of social activities
and intergenerational-only participants having the fewest. Significant psychological well-being
differences were not observed for participants with different age integration statuses, F(2, 55) =
2.56, p = 0.08.
Ordinary Least Squares Regression Analyses
Table 2 presents results from two ordinary least squares regression analyses conducted to
determine the association between age integration status and well-being. The first model
regressed age integration status on psychological well-being while controlling for covariates.
When controlling for sociodemographic and health characteristics, age integration status was not
associated with psychological well-being. Although the independent variable was not
significantly associated with psychological well-being, several covariates were significant.

44
Specifically, Black, non-Hispanic participants were predicted to have psychological well-being
0.81 units greater than their White, non-Hispanic counterparts; those with more than a high
school education were predicted to have psychological well-being 0.72 units greater than those
with a high school education or less; those who received self-care help were predicted to have
psychological well-being 1.32 lower than those who did not receive self-care help; and those
who had an additional health condition were predicted to have psychological well-being 0.41
lower.
In the second model, number of social activities was regressed on age integration status.
Holding all other variables constant, age integration status was associated with social
engagement, such that intergenerational-only participants participated in 0.30 fewer social
activities than their age-integrated counterparts. In addition, women were predicted to participate
in 0.32 more social activities than men, Hispanic and other-raced participants were predicted to
participate in respectively, 0.46 and 0.38 fewer social activities compared to White, non-
Hispanic participants, those with more than a high school education were predicted to participate
in 0.58 more social activities than those with a high school education or less, and those with self-
care help and more chronic conditions were predicted to have 0.23 and 0.12 fewer social
activities respectively, controlling for all other variables.
Discussion
The present study contributes to our understanding of how older adults’ social networks
relate to psychosocial outcomes. By characterizing the age integration status of older adults
based on the presence of intergenerational-only connections, peer-only connections, or both, we
expanded upon prior research that defined age integration as only having intergenerational
connections. Given the well-documented associations of social integration (i.e., diverse social

45
ties with varied social experiences) with mental well-being and social activity participation, we
hypothesized that age-integrated social networks would similarly benefit older adults’ well-being
and social engagement due to the diverse social connections experienced by those with both
same-aged peers and intergenerational connections. This hypothesis was partially supported –
age-integrated participants had significantly greater social engagement than those with
intergenerational-only networks, but not those with peer-only networks; age integration status
was not associated with well-being.
Although the current study did not evaluate the mechanisms through which age-
integrated networks increase social engagement, the hypothesis that having intergenerational and
same-aged peer connections is associated with more social engagement than those with
intergenerational-only networks was confirmed. Given that age-integrated networks inherently
include different types of relationships, it is reasonable to assume that the different relationship
types are associated with different social activities. For example, older adults may participate in
clubs, classes, or organized activities where they engage with same-aged peers; or they may visit
friends and family where they engage with same-aged peers. Age-integrated individuals may
participate in more social activities simply because their varied social connections afford them
greater access to a range of social activities. Given that those age-integrated networks had greater
social engagement than those with intergenerational-only networks, but not those with peer-only
networks, it can be deduced that same-aged peers provide more opportunities for social
engagement than intergenerational connections.
Social engagement was also related to sociodemographic features, including gender, race,
education, and health. As previously documented (Pristavec, 2018), women participated in more
social activities than men. Similarly, those with greater education participated in more social

46
activities than their less-educated counterparts (Pristavec, 2018). We also found that Hispanic
and other-raced older adults were less socially engaged than White, non-Hispanic older adults,
which may be related to cultural differences whereby Hispanic and Asian older adults’
collectivist cultures emphasize familial activities that may take the place of formal social
participation (Guo et al., 2015). Having more health conditions and receiving self-care help were
also associated with fewer social activities, as health limitations have been posited to be both a
precursor to and outcome of disengagement from social activities (Luo et al., 2020; Tomioka et
al., 2017).
Well-being was not related to age integration status as hypothesized; however, several
covariates, including race, education, and health were significantly related to well-being.
Additional analyses investigated the potential moderating effects of race, education, and health in
the association between age integration status and well-being, as these features have been linked
to age integration status and well-being. None of the interactions was significant, suggesting that
sociodemographic features are more strongly associated with well-being than age integration
status. Specifically, Black, non-Hispanic participants had greater well-being that White, non-
Hispanic participants. This finding has been replicated in other studies that have documented a
racial paradox of well-being (Tang et al., 2019), as the cumulative disadvantage experienced by
older Black individuals over their lifetime does not translate into having lower well-being or
poorer mental health than older White individuals with more advantage. Also replicated in other
studies is the finding that more educated individuals have greater well-being (Pinquart &
Sörensen, 2000). This has been explained by their greater access to resources, social capital, and
diverse social ties (Ajrouch et al., 2005). Finally, poorer health, as indicated by having more

47
chronic conditions and especially having help with self-care activities, was associated with lower
well-being, a finding that has also been demonstrated in previous research (Li & Zhang, 2015).
Limitations and Future Directions

Although the current study advanced our understanding of the psychosocial benefits
associated with age integration status, there are several limitations that should be noted. First, the
cross-sectional nature of the data set precludes the causal nature of the relationship between age
integration status and social activities. For example, it is possible that participating in more social
activities leads to richer social networks that include both same-aged peers and intergenerational
connections. At the same time, it is possible that having both same-aged peers and
intergenerational connections opens up opportunities to participate in more social activities with
these social network members. It is likely that both statements are true to some extent; however,
this cannot be determined using cross-sectional data. Supplementary analyses in Appendix B
support the possibility that greater social engagement predicts lower odds of having
intergenerational-only compared to age-integrated networks; however, there remains no
significant association between well-being and age integration status when reversing.
Second, participants were excluded if they did not provide ages for any of their social
network members. This does not mean that those who were excluded from the analysis did not
have social networks; rather, the lack of age information prevented us from determining whether
social network members were intergenerational or same-aged peers. In some cases, participants
remained in the sample because at least one of their social network members had age
information, resulting in a less accurate representation of their complete social network. It is
therefore possible that these social ties that could not be included in the present study have a
significant influence on well-being and social engagement patterns that could not be captured.

48
Although it is not known why age information was missing for some of the participants’ social
network members, it is possible that these individuals were more peripheral social network
members whose ages the participants simply did not know, as they might know their child’s or
spouse’s ages. Social network literature suggests that peripheral, or weak, social ties are strong
predictors of well-being and social engagement, as they provide opportunities for engagement in
diverse settings (Fingerman et al., 2020; Huxhold et al., 2020).
It should also be noted that prior studies assessing well-being and social engagement vary
in operationalization of well-being and social engagement. For example, other studies consider
absence of depressive symptoms or positive self-rated health to indicate psychological well-
being (Fiori et al., 2006; Joiner et al., 2018; Ryff et al., 2015). The current study defines well-
being as a distinct construct made up of eight items that capture subjective, psychological well-
being better than simpler constructs available in NHATS, such as the two-item depressive
symptoms assessment and one-item self-rated health assessment. In addition, the current study
considered the quantity of social activities, but information on the frequency with which older
adults participate in these activities was not available. It is possible that individuals who
participate more frequently in fewer social activities may have greater well-being benefits than
those who participate less frequently in more social activities, because they identify more
strongly with their social roles, which provide them with more meaning and purpose than
activities in which they are less committed (Adams et al., 2011; Baker et al., 2005; Thoits, 2011).
It is also possible that another feature that has not been identified, such as one’s personality, is
driving the association between age-integrated social networks and greater social engagement.
For example, it may be that individuals with certain personality types (e.g., extroverted) are

49
predisposed to having both age-integrated social networks and greater social engagement, in
which case, the association between age integration status and social engagement is confounded.
Even with these limitations, the current study provides new insight on social network
benefits by distinguishing age-integrated networks from intergenerational-only and same-aged
peer networks. Although age integration status was not associated with well-being benefits, it
was associated with social engagement, with age-integrated individuals participating in more
social activities than those with intergenerational-only networks. This demonstrates the value of
distinguishing those with only intergenerational connections from those with intergenerational
connections and same-aged peers. Future research can elucidate the causal nature of this
relationship with longitudinal data.

50
Table 2-1. Weighted Descriptive Statistics
All Participants

N = 3,564
M (SE) or %
Dependent Variables
Well-Being (10-32) 26.97 (0.08)
Social Engagement (0-5) 3.02 (0.03)
Independent Variable
Age Integration Status
Peer 54.76
Intergenerational 20.72
Age-Integrated 24.52
Covariates
Age (67-101) 75.95 (0.12)
Female 52.00
Race/Ethnicity
White, non-Hispanic 83.38
Black, non-Hispanic 6.12
Hispanic 6.82
Other (American Indian/Asian/Native Hawaiian) 3.68
More than HS Education 62.10
Self-Care Help 16.14
Number of Health Conditions (0-9) 2.40 (0.03)

51
Table 2-2. Weighted Descriptive Statistics Stratified by Age Integration Status
Peer-Only

N = 1,768
M (SE) or %
Intergenerational-Only

N = 890
M (SE) or %
Age-Integrated

N = 906
M (SE) or %
Dependent Variables
Well-Being (10-32) 27.09 (0.11) 26.65 (0.17) 26.99 (0.12)
Social Engagement (0-5) 3.03 (0.04) 2.85 (0.07) 3.15 (0.04)
Covariates
Age (67-101) 74.50 (0.14) 79.03 (0.39) 76.60 (0.29)
Female* 46.22 65.03 53.89
Race/Ethnicity**
White, non-Hispanic 85.23 77.16 84.50
Black, non-Hispanic 5.91 8.04 4.98
Hispanic 6.08 9.29 6.38
Other 2.78 5.51 4.15
More than HS Education 62.18 60.57 63.20
Self-Care Help 16.27 13.99 17.66
Number of Health Conditions (0-9) 2.35 (0.04) 2.52 (0.05) 2.39 (0.06)
Note. * p < 0.05, ** p < 0.01, *** p < 0.001.

52
Table 2-3. Weighted Ordinary Least Squares Regression: Effect of Age Integration Status on
Well-Being and Social Engagement
Well-Being

Coefficient (SE)
Social Engagement

Coefficient (SE)
Independent Variable
Age Integration Status
Peer-Only 0.07 (0.16) -0.10 (0.06)
Intergenerational-Only -0.33 (0.21) -0.30 (0.07)***
Covariates
Age (67-101) -0.01 (0.01) 0.00 (0.00)
Female 0.01 (0.14) 0.32 (0.05)***
Race/Ethnicity
Black, non-Hispanic 0.81 (0.17)*** -0.06 (0.07)
Hispanic 0.48 (0.32) -0.46 (0.11)***
Other 0.54 (0.36) -0.38 (0.15)*
More than HS Education 0.72 (0.15)*** 0.58 (0.05)***
Self-Care Help -1.32 (0.20)*** -0.24 (0.07)**
Number of Health Conditions (0-9) -0.41 (0.05)*** -0.12 (0.02)***
Note. Reference categories: age-integrated; male; White, non-Hispanic; high school education or
less; no self-care help. * p < 0.05, ** p < 0.01, *** p < 0.001.

53
CHAPTER 4: HOW AND WHY DOES AGE INTEGRATION STATUS CHANGE OVER
TIME? EXPLORING PERSONAL CHARACTERISTICS AND MAJOR LIFE EVENTS

Introduction
Social convoys, made up of social network connections, are shaped throughout one’s life
course in response to life events (e.g., marriage, education) and individual characteristics (e.g.,
age, race) (Antonucci & Akiyama, 1987). Social ties can be characterized by closeness, such that
relationships in the “inner circle” of one’s convoy represent close emotional others, who tend to
be family and spouses, and relationships in the “outer circle” of one’s social convoy represent
less-close social ties, who tend to be coworkers and acquaintances (Antonucci & Akiyama,
1987). Inner social network members are usually more stable over time than more peripheral ties,
which tend to be influenced more by life circumstances (e.g., whether one is working or
participating in social events). Regardless of their closeness, these social relationships vary
depending on situational and personal characteristics that determine social network functions,
like social support exchanges (e.g., providing and/or receiving physical, emotional, and/or
financial assistance), and social network structures, like size, composition, and social network
type (e.g., family-focused social networks, restricted social networks) (Antonucci & Akiyama,
1987; Berkman et al., 2000).
According to Social Convoy Theory and empirical investigations of older adults’ social
networks, major life changes, like health decline, spousal loss, and retirement, and
sociodemographic features (i.e., age, gender, race, education) are associated with social networks
characteristics and whether they change over time (Antonucci et al., 2019; Cornwell, 2011;
Cornwell & Laumann, 2018; Wrzus et al., 2013). It is important to understand the ways in which
life factors shape social ties over time, as social network characteristics and social network

54
changes contribute to health and well-being through different pathways based on their structural
and functional characteristics (Berkman et al., 2000). For example, it has been documented that
losing a spouse by divorce or widowhood (Cornwell et al., 2008), declining in health, (Cornwell
et al., 2009; Schwartz & Litwin, 2017), and retiring have been associated with social networks
changing in structure and function, typically in ways that reflect loss of social ties and decreasing
social integration (Schwartz & Litwin, 2017).
The current study adds to the literature by exploring whether these findings extend to a
novel social network feature, age integration status, which characterizes participants’ social
networks as intergenerational-only if they include only social network members in a younger
generation (at least 25 years younger), peer-only if they include only connections with same-
aged peers (within 25 years of their age), or age-integrated if they include both types of
connections. Based on prior literature suggesting social network changes in response to spousal
loss, retirement, and health decline (either of older adults or social network members for whom
they provide care), we investigate their associations with age integration status changes. We
posit that age integration status changes are more common for those who experience the major
life events mentioned, while age integration status remains stable for those who do not
experience them. In addition, we explore the sociodemographic (e.g., age, gender, race,
education, marital status) and psychosocial (e.g., well-being, social engagement) predictors of
age integration status transitions, as these features, including how some of them change (i.e.,
getting older, reduced well-being or social engagement) have been linked to social network
features and how they change.
Social Network Correlates

55
Theoretical and empirical research suggests that social ties enhance older adults’ health
and well-being through various mechanisms (Berkman et al., 2000). These include social
support, which refers to instrumental, emotional, and financial support; social engagement,
which reinforces self-worth through social roles and allows for mental and physical stimulation;
social influence, which refers to encouragement or discouragement of health-related behaviors
(e.g., going to the doctor, exercising, eating healthy, drinking, smoking and social comparison;
and greater access to resources (e.g., housing, jobs) through connections. Social network
features, like number and types of social ties, influence the social contexts in which
psychological processes influence health directly (e.g., stress responses, disease transmission, not
smoking), and indirectly (e.g., psychological well-being, self-efficacy, coping strategies).
Role theory suggests that socially integrated older adults with more diverse social ties
have better mental and physical health outcomes than their less socially integrated counterparts
due to the fulfillment of more social roles (Fiori et al., 2006; Litwin et al., 2020; Litwin &
Shiovitz-Ezra, 2011b). The varied social roles allow older adults to engage with others in
different roles that tap into different pathways through which older adults experience cognitive,
physical, emotional, and social stimulation (Fiori et al., 2006). For example, a socially integrated
older woman who identifies as a spouse, mother, daughter, sister, friend, churchgoer, worker,
and volunteer experiences boosted self-efficacy and self-worth from each meaningful social role,
while also experiencing more opportunities to derive psychosocial benefits through multiple
mechanisms than someone who is less socially integrated. That is, one may benefit through
social influence as a friend, social support as a sibling, access to resources as a spouse, social
engagement as churchgoer, reinforced self-efficacy as a worker, and generativity as a volunteer,

56
which might contribute to more positive outcomes compared to a less socially integrated
individual with fewer and less diverse social ties.
It is important to consider the complex relationships among social networks and
downstream factors, like psychosocial mechanisms (e.g., social support, social engagement,
social influence, access to resources) which have positive or negative consequences for health,
through health behaviors, psychological states, and physiological processes. Although this
process suggests causality in one direction (i.e., social networks lead to health), downstream
factors may also influence networks, such that poorer mental or physical health limits one’s
ability to be socially integrated with others, while better mental and physical health encourages
greater social integration. Using the earlier example of a socially integrated older woman with
diverse social roles, we suggested that having more roles reinforces her self-worth and self-
esteem, and in some roles, stimulates physical activity, to benefit her health. It is also possible
that, because of her greater self-worth and self-esteem she is more inclined to take on additional
social roles, and in some cases, because of her greater physical activity abilities, she can socially
engage with others.
Researchers have found support for the bidirectional nature of the social network
associations with mental and physical well-being. For example, Schwartz & Litwin (2019) found
a reciprocal relationship between social integration and mental health, such that better mental
health at baseline led to positive changes in social integration, while greater social integration at
baseline also predicted positive changes in mental health. When comparing the strength of the
effects, they found that mental health was a stronger predictor of social network changes than the
alternative in which social connections influence mental health changes. In the current study, we
focus on age integration status transitions as a result of major life events, such as spousal loss,

57
retirement, health decline, and caregiving responsibilities, that are likely to precede, rather than
result from, social network changes.
Life experiences, including personal characteristics (e.g., age, gender, race, education,
health) and major life events (e.g., parenthood, widowhood, retirement, and health decline),
heavily influence the social relationships individuals form, as they expose individuals to different
social contexts through which they gain social roles. There is a plethora of research on age-
related changes in social networks, with cross-sectional studies documenting that older adults
have smaller social networks than younger adults, and longitudinal studies demonstrating that
older adults’ social networks become smaller with age. In addition to Social Convoy Theory and
empirical investigations that describe how social networks change with age, there are other
supporting theories which attempt to explain age-related social network changes, such as the
frequently observed shrinkage of social networks. For example, according to Disengagement
Theory (Cumming & Henry, 1961), older adults withdraw from social activities as they get older
and experience replacement in societal roles by younger individuals. Socioemotional Selectivity
Theory (Carstensen, 1992) suggests the “pruning” of social network members is a conscious
process in which older adults reduce their focus on less close social ties in order to prioritize
fewer, but more meaningful relationships (English & Carstensen, 2014).
Individuals who vary in sociodemographic, psychosocial, and health characteristics, have
been shown to have distinct social networks across a variety of social network measurements. In
addition to size, social networks change with age by becoming more family-focused, and less
friend-focused (Antonucci et al., 2014; Cornwell et al., 2008; Schwartz & Litwin, 2018; Suanet
& Huxhold, 2020). Recent investigations suggest that later cohorts of older adults may have
larger, and more diverse social networks with age compared to earlier cohorts, which researchers

58
suggest reflects a societal and cultural shift toward maintaining and gaining social connections in
late life (Suanet & Huxhold, 2020). Indeed, social networks have been found to be dynamic with
age (Cornwell et al., 2014, 2021; Litwin et al., 2020; Schwartz & Litwin, 2017). For example,
over a 5-year period, almost all older adults (93%) in the National Social Life, Health, and Aging
Project (NSHAP) lost or gained at least one close social tie, with most experiencing both gains
and losses (Cornwell et al., 2014). Overall, participants added social ties more often than they
removed them; however, the new and removed social connections were not as close emotionally
nor contacted as frequently, as the stable social connections maintained over time (Cornwell et
al., 2014).
Other sociodemographic features have been linked to social network characteristics in
late life, including gender (Antonucci et al., 2014; Schwartz & Litwin, 2018). Women tend to
live longer than men, which may affect their marital status and living arrangements in later life,
leaving them more susceptible to social isolation. At the same time, the social roles that women
uphold throughout their lives as kin-keepers, afford them with greater social support, and more
non-familial social ties, which may buffer against their increased risk of social network reduction
due to living alone or being widowed (Schwartz & Litwin, 2018). Findings support that women
are more socially connected than men, as they have been shown to have larger and more diverse
social convoys that grow over time (Antonucci et al., 2019; Schwartz & Litwin, 2018, 2019).
Researchers have also identified race differences in social networks and how they change
over time (Ajrouch et al., 2001; Nguyen, 2017). Although Black older adults have reported fewer
social network members than White older adults, they have more frequent contact with their
social networks, which consist of more family members (Ajrouch et al., 2001). Researchers have
suggested that racial minorities, and especially Black older adults, experience cumulative

59
disadvantage from a lifetime of discrimination and systemic inequities that place them in a
“socially disadvantaged” environment in which social ties are less stable (Cornwell, 2015).
Based on their assumption that individuals maintain social connections with those from similar
racial backgrounds in socially disadvantaged contexts, in which Black older adults
disproportionately experience poor health outcomes compared to White older adults, they
suggest that social networks may be unstable because social network members have greater risk
of mortality (Cornwell, 2015). The same study found that Latino older adults lost fewer social
network members to death than White older adults, which may reflect a Latino health paradox, in
which Latinos have more health conditions but live longer that Whites, and therefore, remain in
individuals’ social networks over time. Again, this is assuming racial homophily in social
networks, such that Latino older adults primarily include other Latinos in their social network.
An emphasis on communalism and familism may also explain why racial minorities, especially
Hispanic and Asian older adults, focus on social ties within their families (Nguyen, 2017).
Education is a socioeconomic resource that furthers one’s ability to fulfill social,
emotional, and financial needs. As such, it has been suggested that education is associated with
greater social integration because it comes with more opportunities to engage with diverse
individuals in different social roles and greater ability to invest time and resources into
maintaining diverse social ties (Ajrouch et al., 2005; Huxhold et al., 2013; Suanet et al., 2013).
Older adults with lower education have greater difficulty compensating for social losses than
those with greater education (Cornwell, 2015). One study found that later cohorts of older adults
have larger social networks, partially because they had greater educational achievements than
earlier cohorts (Suanet & Huxhold, 2020). Others suggest that the social disadvantage associated
with lower education position one’s social environment to be less stable (Cornwell, 2015). This

60
makes less-educated individuals more likely to experience network changes, and in particular
social network member loss, as those with whom less-educated individuals socially engage may
face similar socioeconomic challenges that put them at greater risk of mortality (Cornwell,
2015).
As previously mentioned, mental health and social engagement are not only outcomes
associated with social network features, but also, can shape social networks and how they
change. For example, Li & Zhang, (2015) found that social network types defined as diverse and
friend, were associated with better mental health than social network types defined as restricted
and family. Given that mental health may change in response to everyday events and life
circumstances (Morack et al., 2013; Zautra et al., 1990), individuals may withdraw from social
ties when they are experiencing depressive symptoms, anxiety, or low well-being, which can
further diminish social engagement opportunities that might buffer against poor mental health
(Howard Litwin et al., 2020; Schwartz & Litwin, 2019). In addition, recent studies have found
that older adults who have experienced loss of social network members or loss of social roles
may choose to become more involved in social activities in order to replenish social network ties
(Cornwell et al., 2014, 2021).
Life Events & Social Networks
In addition to the sociodemographic and health correlates of social network features and
how they change over time, there are several major life events that contribute to changes in
social networks (Cornwell et al., 2021; Litwin et al., 2020; Schwartz & Litwin, 2017).
Researchers have identified changes in structural and functional features of social network in
response to late life transitions, including retirement, spousal loss (i.e., widowhood and divorce),
health decline, and caregiving (Cornwell et al., 2009, 2021; Cornwell & Laumann, 2015; Wrzus

61
et al., 2013). Compared to their counterparts who have not experienced major changes in life
circumstances, older adults who have experienced these transitions are more likely to have social
network changes over time.
Before exploring how life transitions predict social network changes, it is first important
to understand whether and how social networks change over time. Several studies have identified
social network patterns over time, including a longitudinal study over 7 years showing stable
social networks for 34.5% of the sample, and social network changes for 60% of the sample who
transitioned from diverse (i.e., high social integration) networks to family-focused and restricted
social networks (Li & Zhang, 2015). Another study documented similar social network
transitions, with 32% of participants’ social networks remaining the same over 4 years, 37%
increasing in size, and 31% decreasing in size (Schwartz & Litwin, 2018). Although close social
ties tend to remain stable through one’s life, even these social connections are subject to change.
Over 5 years, most older adults lost or gained at least one emotionally close social network
member (Cornwell et al., 2014), while maintaining other relationships and compensating for
losses by replenishing social network members.
Older adults who lose a spouse, whether due to divorce or widowhood, experience major
social network changes (Wrzus et al., 2013). Not only do individuals lose a core member of their
social convoys, but also, they experience shifting dynamics within their social networks (van
Tilburg et al., 2019; Wrzus et al., 2013). For example, bereaved older adults may rely more on
family members or withdraw from social ties (Wrzus et al., 2013). Spousal has been found to
diminish social network quality and quantity, triggering loneliness and social isolation (Pinquart,
2003). Divorced older adults, on the other hand, may rely more on family, or experience
dissolution of previously shared social connections (Wrzus et al., 2013). For example, divorce

62
may include tension with other social network members (e.g., children) and typically eliminates
social ties from the former partner’s social convoy, like in-laws (Wrzus et al., 2013). Recent
findings suggest that losing social network members to death or divorce does not predict
withdrawal from social connections, but instead leads social network changes in which older
adults “cultivate” social ties (Cornwell et al., 2014; Cornwell & Laumann, 2018).
Similarly, physical health limitations affect older adults’ abilities to maintain social
connections and participate in social activities (Cornwell et al., 2009, 2021; Li & Zhang, 2015;
Litwin, 2001; Schwartz & Litwin, 2019). Social settings may not accommodate older adults’
health needs and they may not want or be able to socially engage with others (Cornwell et al.,
2009; Li & Zhang, 2015). Conversely, it is possible that having physical health issues does not
deteriorate one’s social network, but rather, brings together a support system with more social
network members or strengthened social ties who can provide needed instrumental and
emotional assistance (Ayalon & Levkovich, 2019). Chronic health conditions and functional
limitations create social environments that may limit older adults’ social networks in some ways
(e.g., fewer peripheral ties) and strengthen them in others (e.g., greater social support), but
regardless of the mechanisms, there is consensus that health decline shifts social network
features (Schwartz & Litwin, 2019)
Health decline of a close social network member may also be related to social network
changes, not only because that social ties functional and structural contribution to the social
network changes, but also because taking on the caregiver role has been associated with social
network changes (Roth, 2020). Older adults who become caregivers may face social network
changes because their increased time and energy spent on caregiving responsibilities jeopardize
their abilities to social interact with other network members (Roth, 2020). It is also possible that

63
social networks may be strengthened during this time, as they adapt to their new role and
maintain or strengthen social ties in ways that accommodate caregivers’ social constraints, as
older adults’ have faced social network member loss with independent and simultaneous social
network gains (Cornwell & Laumann, 2018). Over a five-year period, older adults who became
caregivers had higher rates of social network transitions, both for social loss and social gains,
than non-caregivers, while those who remained caregivers over 5 years or stopped being
caregivers did not have significant changes in their social networks (Roth, 2020).
Retirement is another life transition that has been linked to changes in social networks, as
individuals experience removal of social environments and social role identities (Schwartz &
Litwin, 2019). There is an inherent change in social networks with retirement, as there are no
longer shared environments for social interactions with more peripheral network members, like
coworkers. For some retired individuals, no longer working allows them to put more time and
energy into strengthening familial connections and engaging in leisure and productive activities.
For other individuals, especially men, social networks shrink in response to retirement (Cornwell
et al., 2008) and may begin a process of social disengagement.
The current study contributes to the literature on how and why older adults’ social
networks change or remain stable over time. Specifically, we offer a unique perspective on social
integration through examination of age integration status, a social network feature which
describes the age composition of older adults’ social networks. Just as socially integrated
networks reflect diverse social ties, age-integrated social networks include age-diverse social
ties, with at least one intergenerational and one same-aged peer connection. Compared to older
adults with peer-only and intergenerational-only networks, those with age-integrated networks
have opportunities to fulfill more varied social roles. We use longitudinal data to explore age

64
integration status transitions over time, including whether life events that predict changes in
social network features similarly predict changes in age integration status. Based on Social
Convoy Theory, which suggests relatively stable close social networks, we predict that age
integration status transitions will be less likely among those who have not experienced major life
events, as they do not experience major social role changes. Given that social network changes
result from major life events like spousal loss, retirement, health decline, and caregiving, it is
seemingly as likely for social networks to change (albeit, in potentially different ways) for those
who experience major life transitions, in which social roles and opportunities shift the other
direction - that is, entering the work force, improvement in health or reduction in functional
limitations, and stopping caregiving. Without specifying exactly how the age integration statuses
change, we hypothesize that those who experience any major life transition, whether positive or
negative, over a 6-year period will be more likely to experience age integration status transitions
as a reflection of their shifting social roles.
Social Convoy Theory and empirical evaluations of social networks over time also
suggest that sociodemographic and health characteristics predispose some older adults to have
social network changes (Antonucci et al., 2014, 2019; Cornwell et al., 2014). Whether stable
(i.e., gender, race, education) or dynamic (i.e., age, well-being, social engagement), personal and
situational features that are not considered major life events also contribute to social network
changes, and as such, may be related to age integration status transitions. Although women may
be more likely to be widowed over the 6-year period, we hypothesize that, when controlling for
marital status, their age integration status will be less likely to change than men’s because they
may be more motivated by social norms to maintain social ties and stability within social
networks. We also predict that Black, non-Hispanic older adults will be more likely than White,

65
non-Hispanic older adults to experience age integration status transitions because health
disparities contribute to unstable social environments, in which they and their social network
members face more limited in how they can engage socially due to environmental or health
constraints and greater risk of mortality. Individuals with lower education are also susceptible to
less stable social environments, which may contribute to greater likelihood of age integration
status transitions. Finally, those with psychosocial health changes over time may be more likely
to experience age integration status transitions; although reduced well-being and social
engagement might lead to age integration status changes through withdrawal from social roles,
boosts in well-being encourage individuals to seek out social ties, and greater social engagement
provides social settings for social network changes, like age integration status transitions.
Methods
Participants and Procedure
Data come from the National Health and Aging Trends Study (NHATS) and are limited
to include participants with age information for at least one social network member, as this
information is necessary to derive age integration status. NHATS is a nationally representative
study of Medicare beneficiaries re-assessed annually to track changes in older adults’ life
experiences, health, and well-being. The current data come from the first round of NHATS data
collection conducted in 2011 (i.e., baseline), and the seventh round of data collection conducted
in 2017 (i.e., follow-up). Sample weights from baseline were used in all analyses to adjust for
nonresponse biases at baseline and attrition over time.
Participants who were present at baseline only (N = 4,914) and follow-up only (N =
2,805) were excluded from the full sample (N = 10,467). Logistic regression analyses were
conducted to determine whether participants present at baseline, but not follow-up, varied in

66
sociodemographic, psychosocial, and health characteristics from those present at both waves (N
= 2,748). Findings suggest that participants who were older and who had intergenerational-only
networks compared to age-integrated networks were more likely to be absent at follow-up. In
addition, Black, non-Hispanic and Hispanic participants were more likely to be absent at follow-
up than White, non-Hispanic participants, while widowed compared to married participants,
more educated participants, and participants with greater well-being and more social activities
were less likely to be present only at baseline. Participants (N = 1,264) were excluded from the
sample due to missing data on sociodemographic (i.e., age, gender, race, marital status,
education), psychosocial (i.e., social engagement, well-being, work status, caregiver status), and
health (i.e., self-care help, chronic conditions) characteristics at either baseline, follow-up, or
both, and if they were never married (due to small sample size). This resulted in a sample of
1,484 participants.
Measures
Age integration status. Information about participants’ age differences with their social
network members was used to designate older adults’ social networks as peer-only (coded as 1)
if they only had social ties at most 25 years older or younger than them, intergenerational-only
(coded as 2) if they only had social ties more than 25 years older or younger than them, or age-
integrated (coded as 3) if they had both peer and intergenerational social ties. Participants could
list up to five social network members, but only those with age information necessary to indicate
their intergenerational or peer nature were included in participants’ social networks. If all social
network members missing age information, participants’ age integration status could not be
derived, and they were excluded from analyses. The use of 25 years as a generational cut-off was
based on research showing that is the length of a generation (Carlsson & Karlsson, 1970). Age

67
integration status patterns over time were used to construct a dichotomous variable, called age
integration status transition, that indicated whether participants experienced any change in age
integration status over time (coded as 1) or whether they remained stable (coded as 0), and age-
integrated changes, a three-category variable indicating whether participants who were age-
integrated at baseline were age-integrated at follow-up (coded as 1), peer-only at follow-up
(coded as 2), or intergenerational-only (coded as 3) at follow-up.
Covariates. Sociodemographic, health, and psychosocial characteristics were included
because of their associations with social network changes and age integration status.
Sociodemographic features assessed at baseline only include gender (coded as 0 = male, 1 =
female), race (coded as 1 = White, non-Hispanic, 2 = Black, non-Hispanic, 3 = Hispanic, 4 =
“other” race/ethnicity), and education (coded as 0 = high school education or less, 1 = more than
high school education). Additional characteristics collected at baseline and follow-up include age
in years, marital status (coded as 1 = married/living with partner, 2 = divorced/separated, 3 =
widowed), labor force status (coded as 0 = not working for pay, 1 = working for pay), whether
participants provide care to another person (coded as 0 = non-caregiver, 1 = caregiver), whether
participants received help with self-care activities in the past year (coded as 0 = no self-care help,
1 = self-care help), number of chronic health conditions. Social engagement was measured at
both time points by counting the number of social activities older adults participated in during
the previous month, including: 1) visiting in person with friends or family; 2) attending religious
services; 3) participating in clubs, classes, or other organized activities; 4) going out for
enjoyment (e.g., going out to dinner, a movie, to gamble, or to hear music or see a play); 5)
doing volunteer work. Greater social engagement was indicated by participation in a greater
number of social activities. Well-being was measured by summing the scores of 8 items that

68
assess emotional well-being and psychological well-being. The former component is evaluated
by asking participants how often they felt emotions (e.g., cheerful) on a Likert scale from 1 =
never to 5 = everyday, while the latter component is assessed by having participants endorse
items like, “My life has meaning and purpose” on a Likert scale from 1 = agree not at all to 3 =
agree. Items were reverse scored such that higher scores indicated greater psychological well-
being.
Major life transitions. Transition variables were created to indicate major life events,
using covariate information collected at baseline and follow-up. Spousal loss is a dichotomous
variable which indicates whether participants’ marital status remained the same over time (coded
as 0) or changed to divorced or widowed (coded as 1). Self-help care transition patterns were
also dichotomized to indicate whether a participant’s self-care help status remained stable (coded
as 0) or changed over time (coded as 1). The latter category combines those whose functional
limitations increased over time (i.e., received help at follow-up only) and those whose functional
limitations decreased over time (i.e., received help at baseline only) to indicate whether
participants experienced major health changes, including improvement and decline in health.
Health condition changes were calculated by subtracting number of chronic conditions at follow-
up from number of chronic conditions at baseline, such that larger numbers indicate greater
decline in health. Job status changes were dichotomized such that those who did not work for
pay at either time point or worked for pay at both time points had stable job statuses (coded as 0),
while those who retired or started a new job were considered to have experienced job status
changes (coded as 1). Similarly, participants had caregiver status changes (coded as 1) if they
became caregivers at follow-up or were no longer caregivers at follow-up, while those who
remained caregivers or non-caregivers over time had stable caregiver statuses (coded as 0).

69
Analytic Strategy
First, we explore age integration status patterns over time. Then we conduct weighted
descriptive statistics at baseline and follow-up on the full sample. Next, we stratify descriptive
statistics by whether participants experienced age integration status transitions or stability over
time. We use chi-square tests and adjusted Wald tests to evaluate bivariate associations of age
integration status transition/stability with sociodemographic, psychosocial, and health
characteristics, and life transitions. Weighted logistic regression models are then used to examine
whether sociodemographic characteristics at baseline, changes in individual characteristics, or
life transitions, predict having any age integration status transition compared to age integration
status stability. We use a model building procedure, in which we first include baseline
characteristics only as predictors of age integration status transitions; next, we add transition
variables created with baseline and follow-up characteristics to account for major life transitions
and evaluate their influence on age integration status change.
Results
As seen in Table 1, age integration status transitions over time were not common.
Although most participants’ age integration statuses remained stable from baseline to follow-up
(72.68%), the remaining 27.32% represent six age integration status transitions, which we
combine into one group to indicate age integration status transition as a dichotomous variable
(i.e., no transition or transition). Weighted descriptive statistics (i.e., means with standard errors
for continuous variables and percentages for categorical variables) were conducted on the
analytic sample (N = 1,484) at baseline and follow-up, as shown in Table 2. Age integration
status proportions changed slightly over time, with peer-only networks decreasing from 59.11%
to 51.90%, intergenerational-only networks increasing from 12.72% to 19.09%, and age-

70
integrated networks remaining similar at 28.17% at baseline and 29.01% at follow-up.
Participants’ ages ranged from 65-91 at baseline, with an average age of 73.13 years old; ages
ranged from 71 to 101 at follow-up, with an average age of 79.21 years old. There were
approximately equal percentages of men (49.68 %) and women (50.32%). Participants were
predominantly White, non-Hispanic (89.94%) and about 64% of participants had more than a
high school education.
Participants’ well-being was 27.81, on average, at baseline and 27.05 at follow-up, with
scores ranging from 11-32, and they reported averages of 3.33 social activities at baseline and
3.12 social activities at follow-up. At baseline, most participants were married or partnered
(73.16%) followed by widowed participants (19.15%) and divorced/separated participants
(7.69%). These percentages shifted slightly at follow-up, with a smaller proportion of
married/partnered individuals (64.42%), a slightly greater proportion of divorced/separated
participants (8.27%), and a greater proportion of widowed participants (27.32%). Most
participants’ marital statuses remained the same over time (90.66%), with the remaining 9.34%
experiencing spousal loss, either through divorce or widowhood, over the follow-up period. The
proportion of participants who received help with self-care activities increased from 8.90% at
baseline to 17.17% at follow-up. When examining the self-care help transition patterns, most
participants had the same self-care help status over time, while 16.56% experienced transitions.
Health conditions increased slightly over time from an average of 2.25 chronic conditions to
2.63. About 28% of participants worked for pay at baseline, while 17.67% worked for pay at
follow-up. Job status changes were experienced by 14.36% of the sample, with the remaining
individuals maintaining their status as worker or non-worker over time. Finally, about 24% of the
sample was caregivers at baseline with almost 18% reporting providing care to others at follow-

71
up. There were caregiver status changes over time for a quarter of the sample, and stability in
caregiving or not caregiving over time for most of the sample.
Table 3 presents weighted descriptive statistics stratified by whether participants had any
age integration status transition or whether their age integration status remained stable over time.
Participants who had age integration transitions were older, on average, than those who
maintained their age integration status, F(1, 56) = 5.04, p < 0.05. Although there were more
females (52.81%) than males in the stable group, the transition group consisted of a lower
proportion of females (43.72%) than males, χ
2
= 9.74, df = 1, p < 0.01. Race was similar across
groups, χ
2
= 2.16, df = 3, p = 0.475, with majority White, non-Hispanic participants (90.17% and
88.40%), followed by increasingly smaller proportions of participants who were Black, non-
Hispanic, Hispanic, and other races.
Participants in both the transition and stable age integration status groups had similar
average social engagement at baseline, F(1, 56) = 0.17, p = 0.683, and follow-up, F(1, 56) =
0.43, p = 0.515, as well as similar average changes in social engagement, F(1, 56) = 0.09, p =
0.762. Similarly, there were no significant group differences in well-being at baseline, F(1, 56) =
0.10, p = 0.750. or at follow-up, F(1, 56) = 0.16, p = 0.691, or in changes in well-being, F(1, 56)
= 0.68, p = 0.413.
Marital status at baseline, χ
2
= 22.15, df = 2, p < 0.001, was significantly associated with
age integration status transition or stability, while marital status at follow-up was not
significantly related, χ
2
= 4.788, df = 2, p = 0.139. The transition group had a greater majority of
married/partnered participants at baseline (81.94%) than the no transition group (69.86%). At
follow-up, the proportion of married/partnered participants in the transition group was reduced to
61.50%, with the percentage of widowed participants increasing from 13.50% to 31.31%. The

72
stable group experienced these changes over time to a smaller degree: the percentage of
married/partnered participants was reduced to 65.51%; the percentage of divorced/separated
participants remained about the same (8.67% to 7.19%); and the percentage of widowed
participants increased from 21.32% to 25.81%. Overall, 20.55% of the transition group
experienced spousal loss, while a significantly smaller proportion (5.12%) of those in the stable
group lost a spouse from baseline to follow-up, χ
2
= 82.90, df = 1, p < 0.001.
The proportion of participants who received self-care help was similar in both groups at
baseline, χ
2
= 0.67, df = 1, p = 0.442, and at follow-up, χ
2
= 0.01, df = 1, p = 0.904. The self-care
help transition patterns were also similar across groups, χ
2
= 0.00, df = 2, p = 0.998, with most
participants not experiencing self-care help changes. Health conditions were not significantly
different among groups at baseline, F(1, 56) = 3.87, p = 0.054, or follow-up, F(1, 56) = 1.49, p =
0.227. Change in average health conditions was also not significantly different between those in
the transition group and the stable group, F(1, 56) = 1.01, p = 0.320.
Finally, job status at baseline, χ
2
= 0.768, df = 1, p = 0.381, job status at follow-up, χ
2

= 0.36, df = 1, p = 0.571, and transitions in job status, χ
2
= 0.05, df = 1, p = 0.854, were not
significantly different for those who experienced age integration status transitions versus
stability. Similarly, there were no significant differences among groups with respect to changes
in caregiver status over time, χ
2
= 0.98, df = 1, p = 0.358, caregiver status at baseline, χ
2
= 0.93,
df = 1, p = 0.457, or caregiver status at follow-up, χ
2
= 0.00, df = 1, p = 0.964.
Multivariate Results
Table 4 presents results from weighted multivariate logistic regression models predicting
any change in age integration status over time compare to age integration stability. The first
model explores the cross-sectional association of baseline sociodemographic, psychosocial, and

73
health characteristics with age integration status transition over time. Findings show older
participants were more likely to increase age integration status transition such that with each
additional year, their odds of transition increase by 3%. Similarly, with each additional health
condition, controlling for other factors, participants were predicted have a 12% increase in odds
of age integration status transition. Black, non-Hispanic participants also had 1.74 times greater
odds of age integration status transition compared to White, non-Hispanic participants, taking
other variables into account. Compared to those who were married at baseline,
divorced/separated and widowed participants both had 0.44 times lower odds of experiencing an
age integration status transition. The other variables in the model (i.e., gender, education, well-
being, social engagement, self-care help, job status, and caregiver status) were not significantly
related to age integration status transition.
Model 2 explores longitudinal associations by adding changes in sociodemographic and
health characteristics over time to see if life events, like spousal loss, self-care help changes,
health condition changes, job status changes, and caregiver status changes predict transitions in
age integration status compared to stability. Unlike Model 1, age is no longer associated with age
integration status transitions, while gender is now a significant predictor, such that women had
0.71 times lower odds of experiencing age integration status transitions compared to men. Like
Model 1, Black, non-Hispanic participants were more likely to have age integration status
transitions compared to their White, non-Hispanic counterparts. Also similar to Model 1,
findings show that those who were divorced/separated or widowed at baseline had 0.53 times
and 0.66 times lower odds, respectively, of having age integration status transitions than
married/partnered participants. The addition of the spousal loss variable accounts for age
integration status changes, such that those who lost a spouse had 4.45 times greater odds of age

74
integration status transition. The other change variables (i.e., job status change, self-care help
change, health conditions change, caregiver status change) were unrelated to age integration
status transitions, contrary to our hypotheses. Finally, those who were caregivers at baseline had
0.66 times lower odds of experiencing age integration status transitions than non-caregivers.
Discussion
Age integration status was found to be a stable feature of social networks over time for
most older adults. Although most older adults did not experience transitions in age integration
status, those that did were posited to have distinct life circumstances that prompted their social
network changes. Given that social network functions and structures have been shown to change
in response to major life changes, we hypothesized that age integration status as a social network
feature would similarly change as a result of life events. Further, we expanded upon previous
research investigating life transition influences on social network changes by considering
positive and negative life event transitions. That is, rather than only investigate the effect of
health changes on social networks by looking at health decline, we also considered that those
experiencing positive health changes undergo a major life change. When examining work
changes, we considered not only retirement, but also going back to work in late life; similarly,
when examining caregiver status, we included those who became caregivers, and those who
stopped caregiving, as these transitions represent major life changes regardless of the direction of
change (e.g., retirement or going back to work; health decline or health improvement).
Age integration status is a unique social network feature that considers not only whether
individuals have intergenerational connections or not, but also whether intergenerational
connections are accompanied by same-aged peer connections. Prior studies on age integration
are limited because they simply compare presence or absence of intergenerational social network

75
members to indicate age integration. This limits the opportunity to distinguish those who have
both intergenerational and same-aged peer connections from those who have only
intergenerational connections, even though the former inherently have greater age integration
because of their varied, age-diverse social roles. In addition, age integration status allows us to
compare fundamental differences among those with intergenerational-only, age-integrated, and
peer-only networks.
We found that age integration status changes over a 6-year period were not common,
consistent with Social Convoy Theory and studies documenting stability in close social ties over
time (Antonucci & Akiyama, 1987; Antonucci et al., 2014). The proportion of older adults
whose age integration status remained stable (72.68%) was relatively high compared to studies
that have reported social network stability rates around 35% (Li & Zhang, 2015) and 32%
(Schwartz & Litwin, 2018). Age integration status therefore may be a more stable social network
feature than those that have been shown to change over time, like social network type and size
(Li & Zhang, 2015; Schwartz & Litwin, 2018). Of those whose age integration status stayed the
same, over half had peer-only social networks, while 22% and 21% had age-integrated and
intergenerational-only social networks respectively.
It is possible that there are age integration status changes that occur in response to major
life transitions, but they are not captured in the current data. Given the 6-year follow-up period, it
is possible that in between data collection rounds, participants experienced major life events that
led to unobserved age integration status changes followed by adaptations to replenish social ties
that make it seem like participants’ age integration status never changed. It is also possible that
the age integration status changes that we expected to result from major life transitions happened
prior to the start of the study. This is supported by our finding that those who had already

76
experienced spousal loss at baseline were less likely to experience age integration transitions
over the course of the study, because they could no longer lose their deepest social connection. In
addition, the low rates of major life transitions in the current sample may explain why we
observed low rates of age integration status transitions. The 6-year follow-up period may not
have been long enough to observe the changes that we hypothesized because major life
transitions had not yet occurred. For example, risk of health decline and likelihood of becoming
a caregiver increase with age, so participants may still undergo major life changes, and
associated age integration status changes, after the 6-year follow-up period.
Our findings show that gender, race, marital status, spousal loss, and caregiver status are
associated with age integration status changes over 6 years. Women were less likely to
experience age integration status transitions, which is consistent with research showing that due
to a lifetime of gendered social role norms, older women are better equipped and more likely to
maintain social connections than men (Antonucci et al., 2004; Moen, 2001; (Schwartz & Litwin,
2018, 2019)). With more larger networks and more diverse social roles (Schwartz & Litwin,
2018; Schwartz & Litwin, 2019), it is possible that women do experience social network changes
that do not affect age integration status, because they have other intergenerational and/or peer
social ties that “cover the losses” and contribute to their peer-only, intergenerational-only, or
age-integrated social network statuses. Put another way, losing a peer or intergenerational social
tie may not change age integration status if there are other peers and intergenerational
connections; for those who have smaller social networks (e.g., men), loss of a peer or
intergenerational connection may have a greater impact on age integration status.
We found support for our hypothesis regarding race differences in age integration status
transition. Compared to White, non-Hispanic participants, Black, non-Hispanic participants were

77
more likely to have age integration status changes. Although we do not know the nature of these
changes (e.g., whether participants change from peer-only or intergenerational to age-integrated),
the relative lack of stability in age integration status may reflect instability in their social
environments (Cornwell, 2015). It is possible that Black older adults and their social network
members are more likely to have age integration status changes, because their lifetime
experiences of cumulative disadvantage contribute to greater risk of health problems and
mortality that interfere with social connectedness. It is also possible that age integration status is
more susceptible to change for Black older adults as they have smaller social networks (Ajrouch
et al., 2001; AnnW. W. Nguyen, 2017) and therefore, loss of social ties may have a stronger
impact on age integration status than loss of social ties from a larger network. For example,
someone who loses a peer from a small, age-integrated network with one other intergenerational
tie will transition to an intergenerational-only network; on the other hand, someone who loses a
peer from a larger age-integrated network with two peers and two intergenerational ties will
remain age-integrated.
As expected, marital status and spousal loss were associated with age integration status
changes, such that those who were divorced or widowed at baseline were less likely than married
individuals to experience age integration transitions, while those who got divorced or widowed at
follow-up were more likely to experience age integration transitions. Given that
married/partnered older adults tend to be age-integrated, while divorced/separated and widowed
older adults tend to have peer-only and intergenerational-only networks, there is more “room for
change” in married/partnered participants’ social networks. Married/partnered participants with
age-integrated, and even peer-only networks, are at greater risk of experiencing age integration
status changes, because they are susceptible to social network changes not only in response to

78
their own life circumstances, but also their spouses. On the other hand, compared to those whose
marital status does not change (i.e., participants who remain married, divorced/separated, or
widowed over time), those who lost a spouse to divorce or death from baseline to follow-up had
greater likelihood of age integration status transitions. This supports our hypothesis and other
research which shows spousal loss to be a disruptive event leading to major social network
changes (Cornwell et al., 2009; Wrzus et al., 2013), as older adults lose their closest social tie,
and in response may further withdraw from social networks, lose shared social connections, or
compensate by seeking out or strengthening other social ties (Cornwell & Laumann, 2018).
Although we expected change in caregiver status to be a major life change that would
predict age integration status transitions, this was not supported by the current findings; however,
baseline caregiver status was associated with age integration status transitions, such that
caregivers were less likely than non-caregivers to experience any change. It is possible that due
to their role earlier on, controlling for whether they change over time or not, caregivers create
stable social networks that accommodate and support their caregiver responsibilities. Non-
caregivers may have more flexibility to change their social networks.
Several other hypotheses were not supported regarding the influence of major life
changes and personal characteristics on age integration status transitions. First, we expected
participants in poorer health (i.e., had self-care help and more health conditions) and who
experienced health changes (i.e., self-care help transitions and health condition changes) to be
more likely to experience age integration status changes over time. The only association we
observed in support of our hypothesis was in the cross-sectional model, in which those who had
more health conditions at baseline were more likely to have age integration status changes,
possibly due to health-related social limitations (Li & Zhang, 2015). However, in the

79
longitudinal model which also accounts for change in health conditions, neither change nor
baseline health conditions were associated with age integration status transitions. In addition,
age, education, job status and job status changes were not related to social network change,
contrary to our hypotheses.
Limitations and Future Directions
Though the current findings add to the literature by exploring whether and why age
integration status changes over time, there are several limitations to consider. First, we created
the age integration status variable using social network member information that indicates
whether participants are in another generation (i.e., intergenerational) or not (i.e., same-aged
peers). Although we selected 25 years to indicate the span of a generation, this is a subjective
cut-off point that could vary based on individuals’ perceptions of what intergenerational
connections are. For example, some individuals may think that someone 10 years younger than
them is an intergenerational connection, while others may think that someone more than 25 years
younger than them, or below a certain age (i.e., less than 30 years old). In addition, the 25-year
age difference may exclude individuals who are just outside of that range, such that a child 26
years younger than them would be designated a peer connection, while a child 24 years younger
than them would be considered an intergenerational connection, despite them both being the next
generation.
In addition, we did not have large enough sample sizes to explore all age integration
status transition patterns. Although understanding predictors of transitions versus stability is an
important contribution, future research would benefit from examination of specific patterns. For
example, what predicts changing from age-integrated to peer-only, or from intergenerational-
only to age-integrated.). It is possible that the small sample sizes that preclude multivariate

80
analyses of age integration status transition patterns, are evidence of lack of change; however, it
is also possible that age integration status transitions are missed due to measurement issues. That
is, if social network members were missing age information at one or both waves, they were not
considered in the older adult’s social network, which may skew our perception of age integration
statuses that do not match actual experiences of age integration in social networks.
Despite these weaknesses, the present study provides new and useful information on age
integration status changes over time in a nationally representative sample of older adults. We
showed that age integration status was relatively stable over 6 years and that some of the life
changes typically associated with social network changes similarly predict age integration status
changes. Future studies can expand upon this two-wave data analysis by incorporating data from
other NHATS rounds conducted between baseline and follow-up, and after follow-up, as the
study conducts annual reassessments. In addition, future research can explore how age
integration status changes may lead to changes in psychosocial outcomes, and whether the
bidirectional nature of social network features with well-being and social engagement will be
similarly observed for age integration status.

81
Table 3-1. Transition Variables (N = 1,484)
Baseline Follow-Up N Weighted
Percentage
Change Variable
Age Integration Status Age Integration Status Age Integration Status Transition
Peer-Only Peer-Only 599 44.70 Stable
Age-Integrated Age-Integrated 231 16.67 Stable
Intergenerational-Only Intergenerational-Only 222 11.31 Stable
Peer-Only Age-Integrated 171 11.49 Transition
Age-Integrated Peer-Only 97 6.66 Transition
Age-Integrated Intergenerational-Only 82 4.85 Transition
Peer-Only Intergenerational-Only 55 2.93 Transition
Intergenerational-Only Age-Integrated 18 0.86 Transition
Intergenerational-Only Peer-Only 9 0.55 Transition

Marital Status Marital Status Spousal Loss
Married/Partnered Married/Partnered 843 64.42 No Loss
Married/Partnered Divorced/Separated 14 0.79 Loss
Married/Partnered Widowed 141 7.95 Loss
Divorced/Separated Divorced/Separated 115 7.29 No Loss
Divorced/Separated Married 0 0 No Loss
Divorced/Separated Widowed 6 0.40 Loss
Widowed Widowed 362 18.96 No Loss
Widowed Married/Partnered 0 0 No Loss
Widowed Divorced/Separated 3 0.19 Loss

Self-Care Help Status Self-Care Help Status Self-Care Help Change
No Help No Help 1,134 78.68 No Change
Help Help 80 4.76 No Change
Help No Help 58 4.14 Change
No Help Help 212 12.42 Change
Job Status Job Status Job Status Change
Non-Worker Non-Worker 1,086 67.97 No Change
Worker Worker 164 13.90 No Change
Worker Non-Worker 185 14.36 Change
Non-Worker Worker 49 3.77 Change

Caregiver Status Caregiver Status Caregiver Status Change
Non-Caregiver Non-Caregiver 1,021 66.41 No Change
Caregiver Caregiver 111 8.36 No Change
Caregiver Non-Caregiver 226 15.77 Change
Non-Caregiver Caregiver 126 9.46 Change

82

Table 3-2. Weighted Descriptive Statistics, Baseline and Follow-Up (N = 1,484)
Variable (Range at baseline; Range at follow-up) Baseline Follow-Up

M (SD) or %

M (SD) or %
Age Integration Status Transition (ref = no transition) - 27.32
Age Integration Status
Peer-Only 59.11 51.90
Intergenerational-Only 12.72 19.09
Age-Integrated 28.17 29.01
Age (65-95; 71-101) 73.13 (6.10) 79.21 (6.10)
Female 50.32 -
Race/Ethnicity
White, non-Hispanic 89.94 -
Black, non-Hispanic 4.03 -
Hispanic 3.89 -
Other 2.14 -
More than HS Education 64.22 -
Well-Being (12-32; 11-32) 27.81 (2.93) 27.05 (3.21)
Well-Being Change (-16-11) - -0.76 (2.76)
Social Engagement (0-5) 3.33 (1.19) 3.12 (1.26)
Social Engagement Change (-5-4) - -0.21 (1.11)
Marital Status
Married/partnered 73.16 64.42
Divorced/separated 7.69 8.27
Widowed 19.15 27.32
Spousal Loss (ref = same marital status) - 9.34
Self-Care Help 8.90 17.17
Self-Care Help Change (ref = no change) - 16.56
Health Conditions 2.25 (1.47) 2.63 (1.45)
Health Conditions Change - 0.38 (1.12)
Job Status 28.26 17.67
Job Status Change (ref = no change) - 14.36
Caregiver Status 24.13 17.82
Caregiver Status Change (ref = no change) - 25.22
Note. Variables only in 2011 are time-invariant. * p < 0.05, ** p < 0.01, *** p < 0.001.

83

Table 3-3. Weighted Descriptive Statistics by Age Integration Status Transition (N = 1,484)

Stable

N = 1,052
M (SE) or %
Transition

N = 432
M (SE) or %

p
Age 72.93 (6.15) 73.68 (5.96) *
Female 52.81 43.72 *
Race/Ethnicity
White, non-Hispanic 90.51 88.40
Black, non-Hispanic 3.61 5.17
Hispanic 3.74 4.29
Other 2.14 2.14
More than HS Education 64..57 63.28
Well-Being Baseline 27.79 (2.87) 27.86 (3.08)
Well-Being Follow-Up 27.07 (3.16) 26.99 (3.34)
Well-Being Change -0.72 (2.71) -0.87 (2.89)
Social Engagement Baseline 3.33 (1.19) 3.35 (1.17)
Social Engagement Follow-Up 3.11 (1.26) 3.15 (1.25)
Social Engagement Change -0.22 (1.12) -0.20 (1.09)
Marital Status Baseline **
Married/partnered 69.86 81.94
Divorced/separated 8.82 4.66
Widowed 21.32 13.40
Marital Status Follow-Up
Married/partnered 65.51 61.50
Divorced/separated 8.67 7.19
Widowed 25.81 31.31
Spousal Loss (ref = no loss) 5.12 20.55 **
Self-Care Help Baseline 9.27 7.91
Self-Care Help Follow-Up 17.25 16.99
Self-Care Help Change (ref = no change) 16.56 16.57
Health Conditions Baseline 2.20 (1.47) 2.38 (1.44)
Health Conditions Follow-Up 2.60 (1.45) 2.70 (1.44)
Health Conditions Change 0.40 (1.12) 0.32 (1.13)
Job Status Baseline 28.89 26.59
Job Status Follow-Up 18.04 16.70
Job Status Change (ref = no change) 14.24 14.69
Caregiver Status Baseline 27.49 22.39
Caregiver Status Follow-Up 17.78 17.91
Caregiver Status Change (ref = no change) 24.54 27.05
Note. Variables only in 2011 are time-invariant. * p < 0.05, ** p < 0.01, *** p < 0.001.

84
Table 3-4. Logistic Regression Predicting Age Integration Status Transition (N = 1,484)
Age Integration Status Transition
Model 1: Cross-
Sectional
Model 2: Longitudinal
OR [95% CI] OR [95% CI]
Age at Baseline 1.03** [1.01, 1.05] 1.01 [0.99, 1.04]
Female 0.82 [0.62, 1.04] 0.71* [0.53, 0.95]
Race/Ethnicity (ref = White, non-Hispanic)
Black, non-Hispanic 1.74** [1.19, 2.51] 1.68* [1.12, 2.53]
Hispanic 1.30 [0.70, 2.39] 1.36 [0.70, 2.63]
Other 1.02 [0.40, 2.96] 1.10 [0.42, 2.87]
More than HS Education 0.91 [0.69, 1.23] 0.92 [0.70, 1.22]
Well-Being Baseline 1.00 [0.96, 1.05] 1.01 [0.96, 1.06]
Well-Being Change 0.98 [0.94, 1.04]
Social Engagement Baseline 1.03 [0.92, 1.15] 1.09 [0.96, 1.24]
Social Engagement Change 1.05 [0.92, 1.20]
Marital Status Baseline (ref = Married/partnered)
Divorced/separated 0.44** [0.24, 0.78] 0.53* [0.29, 0.98]
Widowed 0.44*** [0.31, 0.64] 0.66* [0.45, 0.97]
Spousal Loss (ref = no loss) 4.45*** [2.99, 6.63]
Self-Care Help (ref = no help) 0.74 [0.48, 1.16] 0.82 [0.51, 1.32]
Self-Care Help Changes (ref = no change) 0.96 [0.67, 1.37]
Health Conditions Baseline 1.12* [1.03, 1.23] 1.11 [1.00, 1.23]
Health Conditions Change 1.01 [0.89, 1.15]
Job Status (ref = no job) 0.95 [0.69, 1.29] 0.84 [0.55, 1.29]
Job Status Changes (ref = no change) 1.22 [0.72, 2.08]
Caregiver Status (ref = non-caregiver) 0.93 [0.68, 1.26] 0.66* [0.45, 0.97]
Caregiver Status Changes (ref = no change) 1.31 [0.90, 1.90]
Note. * p < 0.05, ** p < 0.01, *** p < 0.001.

85
CHAPTER 5: SUMMARY & DISCUSSION

The current evaluation of age integration status in a nationally representative sample of
Medicare beneficiaries in the United States contributes unique information on the age
composition of older adults’ social networks. Given the importance of social connections
(Berkman et al., 2000), and the increasing availability of intergenerational social ties as society
becomes more age-integrated (Riley & Riley, 2000), we suggest that older adults have more
opportunities to have age-integrated social networks which may be associated with unique
psychosocial benefits. We characterized older adults with at least one social network member in
both a younger generation (i.e., at least 25 years younger) and the same generation (i.e., within
25 years of their age) as having age-integrated networks due to the age-diverse nature of their
social networks. We posit that having age-diverse ties indicates presence of age integration in
social networks, much like having diverse social ties indicates social integration in social
networks.
Prior attempts at operationalizing age integration within social networks have compared
older adults with intergenerational ties to those without, while overlooking the contribution of
peers. These studies do not account for the social context in which one experiences
intergenerational connections or not, nor do they consider that true age integration is experienced
when one has not only intergenerational connections, but also social ties with those in their own
generation. Age integration status allows us to compare older adults with age-integrated social
networks (i.e., with both intergenerational and peer ties) to older adults with intergenerational-
only and peer-only networks. We hypothesized that age integration status, much like other social
network features, would be associated with sociodemographic, health, and psychosocial

86
characteristics of older adults. This dissertation furthers our understanding of age integration
status in older adults’ social networks through cross-sectional and longitudinal studies evaluating
which older adults have age-integrated, intergenerational-only, and peer-only networks, whether
age-integrated networks benefit those who have them, and how age integration status changes
over time.
Study 1
The first study adds to the scarce empirical evidence of age integration in older adults’
social networks. Using data from the seventh round of NHATS and guided by Social Convoy
Theory, we found that older adults’ age integration status is related demographic (i.e., age,
gender, race), sociodemographic (i.e., marital status, education), and health (i.e., whether the
participant received self-care help, chronic conditions) characteristics. We used weighted
multinomial logistic regressions to explore age integration status group differences by comparing
the two groups with intergenerational connections (i.e., age-integrated and intergenerational-
only) to those with peer-only networks. Almost half of the participants had peer-only networks
whereas the remaining participants were split between age-integrated (26%) and
intergenerational-only (25%) networks. Older age was associated with greater odds of
intergenerational-only and age-integrated networks compared to peer-only. Women were
predicted to have more age-integrated than peer-only networks than men. No other
sociodemographic features predicted age-integrated networks. Intergenerational-only networks
were more likely than peer-only networks for Hispanic, other-raced, non-partnered, and more
educated participants. Peer-only networks were more likely than intergenerational networks
among participants who received self-care help. Older age and gender correlated with age-
integrated and intergenerational networks, with minority and unpartnered participants belonging

87
to the latter. These results provide a foundation for future studies to evaluate psychosocial
correlates of age integration status. Differences in sociodemographic factors were observed most
often between age-integrated networks and intergenerational-only networks. Older age correlated
with both age-integrated and intergenerational networks. The sociodemographic differences
revealed among older adults with intergenerational-only, peer-only, and age-integrated networks
demonstrates the value in re-conceptualizing and re-operationalizing age integration status and
provides a foundation for studies to evaluate psychosocial correlates of age integration status.
Study 2
The second study examined age integration status associations with psychosocial
outcomes in a cross-sectional evaluation of well-being and social engagement. We hypothesized
that age-integrated older adults would have greater well-being and participate in more social
activities than older adults with peer-only or intergenerational-only networks, due to the
additional social roles age integration affords, and the additional pathways that come with
different relationship through which age-integrated older adults might experience psychosocial
benefits. We found partial support for our hypotheses, as older adults with age-integrated social
networks did not have differ in well-being from those with peer-only or intergenerational-only
networks, but they did have greater social engagement than those with intergenerational-only
networks. The social engagement benefits may be exclusive to those with peers in their social
network, as evidenced by those with intergenerational-only networks having significantly fewer
social activities than age-integrated older adults. It is possible that peers provide greater
opportunities for engagement in varied social activities (e.g., visiting with friends and family,
social outings, and club events) while intergenerational connections facilitate a limited set of
social activities (i.e., visiting with friends and family).

88
There are several reasons why well-being was not associated with age integration status.
First, it is possible that individuals self-select into social networks, such that older adults with
age-integrated social networks have certain personalities or characteristics that encouraged their
age integration status, while those with peer-only and intergenerational-only networks have other
characteristics that allowed them to self-select into their groups. Age integration status, therefore,
may be associated with well-being benefits, not through varied age-diverse social ties as
hypothesized, but through social control that allows older adults to define their own social
networks and maintain social ties that bolster their well-being, whether those are only peers, only
intergenerational connections, or both (Berkman et al., 2000). Second, it is possible that age
integration status excludes more peripheral social ties that are necessary for well-being. This may
be due to older adults not listing weaker social ties when asked to list up five people they would
share important information with older adults, or because listed social network members did not
have age information to contribute to age information status and were excluded from older
adults’ social networks.
The other sociodemographic features found to be associated with both well-being and
social engagement include race, education, and health; gender was associated with social
engagement only. Race differences varied by outcome such that Hispanic and other-raced older
adults had lower social engagement than White, non-Hispanic older adults, while Black, non-
Hispanic participants had greater well-being than White, non-Hispanic older adults. In line with
our hypotheses, those with different education levels had similar group differences for well-
being and social engagement, with more-educated older adults reporting greater social
engagement and well-being than those with a high school education or less. As hypothesized,
and consistent with other research (Luo et al., 2020; Li & Zhang, 2015; Tomioka et al., 2017),

89
those in poorer health had worse psychosocial outcomes than their healthier counterparts. The
second study made important contributions to our understanding of the psychosocial correlates of
age integration status. Although the cross-sectional results do not allow for causal interpretations,
we found support the possibility that social engagement leads to age integration status and
suggest that future studies investigate the bidirectional of age integration status.
Study 3
Study 3 provides insight on whether age integration status is like other social network
features which have been shown to change over time in response to major life events and
personal characteristics that typically lead to changes in social networks. We conducted two-
wave longitudinal data analysis to explore whether age integration status changed or remained
stable over 6 years. We found that most older adults had stable age integration statuses over time,
while about 27% had transitions. Those whose age integration status transitioned displayed 6
different patterns of age integration changes, with group sizes too small to make meaningful
conclusions. Using weighted logistic regression, we predicted how likely older adults were to
experience any age integration status transition over time (compared to no transition) as a
function of sociodemographic and psychosocial characteristics, and major life changes, including
spousal loss, health change, job status change, and caregiver status change. Prior studies have
shown that social networks are fundamentally disrupted by these events because they change
one’s life circumstances and reshape social convoys (Antonucci & Akiyama, 1987; Cornwell et
al., 2021). We did not find support for our hypotheses regarding most of these life changes,
except in the case of spousal loss. Those who got divorced or were widowed in the 6-year
follow-up period were more likely to have age integration status changes, which appropriately
reflects the loss of a focal social network member. If participants were already widowed or

90
divorced at baseline, they were less likely to have age integration status transitions, possibly
because their social networks had already undergone changes in response to spousal loss. Also,
in the cross-sectional model, we found that baseline health conditions predicted transitions in age
integration status, such that those in poorer health were more likely to experience transitions, as
hypothesized. Finally, Black, non-Hispanic older adults were more likely to experience age
integration status transitions, while women and those who were caregivers at baseline were less
likely. The study findings suggest that age integration status stability is more common than
transitions and that major life changes, aside from spousal loss, do not predict changes in age
integration status like they do other social network features. It is possible that one reason we did
not find the associations as expected is the relative lack of major life transitions in the sample. It
is difficult to assess age integration status transitions as a result of major life changes if there are
not many changes to examine.
General Discussion
The three studies in this dissertation provide us with a new lens on age integration within
older adults’ social networks. Reconceptualizing age integration in social networks as having
age-diverse social ties allowed us to compare three groups of older adults with distinct social
experiences depending on their social integration with individuals their own age, in younger
generations, or both. Indeed, we found that these groups were distinct not only in how they
varied in age integration status, but also in demographic, sociodemographic, and psychosocial
features. There are other characteristics that can be explored in relation to age integration status,
like personality and attitudes. Age integration status may reflect personality – for example,
extroverted people may branch out to others outside of their age group, while introverted people
may maintain more comfortable familiar social ties with people their own age. It is possible that

91
age integration status within social networks is influenced by attitudes toward younger
individuals, given that more positive attitudes might encourage more intergenerational
connections according to Age Integration Theory (Riley & Riley, 2000). It is also possible that
intergenerational connections in organized settings, like intergenerational programs that
purposefully connect older and younger individuals for meaningful shared connections and
activities, are important contributors to age integration status that we are unable to evaluate in the
current data set (Jarrott et al., 2018).
Although the current dissertation contributed unique information about age composition
of social networks using a nationally representative data set, there are some methodological
limitations to the operationalization of age integration status within the current data set that
should be noted. First, the age integration status variable was derived from social network
information on up to 5 close social ties. It is possible that age integration status is better
measured by including social network members beyond the inner social circle. There may be
more variability in age integration status than observed if older adults listed more social network
members. Additionally, we could observe age integration status more comprehensively by
including additional information about social ties that is not collected in the current data, like
contact frequency and relationship quality. Similarly, it would be useful to have a data set that
prompts individuals to consider their intergenerational and peer ties or asks participants to
subjectively evaluate what types of relationships they consider to be intergenerational to ensure
that the operationalization of age integration takes into account older adults’ subjective
experiences with age integration.
Also, age integration status combines information from two binary variables – presence
of intergenerational ties and presence of peer ties. Although we have information available on

92
the number of intergenerational ties and number of peer ties, the current data simply consider age
integration status which only indicates whether individuals have at least one of each tie (i.e., age-
integrated) or have at least one or the other (i.e., peer-only or intergenerational-only), and not the
number of each type of tie. Therefore, we miss out on use of information that might further
distinguish individuals’ experiences with social network age integration, like the ratio of
intergenerational to peer ties within social networks. In addition, we were limited by using
available age information on social network members and having to exclude social network
members missing age information. Given that we created age integration status in recognition of
the importance of the full social context (i.e., considering presence or absence of peers in
addition to intergenerational connections), it is important to acknowledge the slight inaccuracy of
age integration status, as it cannot include social network members missing age information, but
that does not mean they do not exist – in contrast, they do exist and may even contribute to age
integration experiences. Therefore, in supplemental analyses, we categorize older adults’ social
network members as “peer,” “intergenerational,” or “missing age” so we could investigate their
contributions to social networks beyond age information – specifically, we use the available data
on relationship types of social networks members to investigate the missing age relationships and
see if additional information on age integration status can be gleaned in spite of missing age
information.
In Appendix C, we present data on all participants at baseline (N = 7,662) and all
participants at follow-up (N = 5,553) to explore types of relationship (i.e., spouses, parents,
siblings, children, grandchildren, other relatives, and non-relatives) connections (i.e.,
intergenerational, peer, and missing age information). Given that age integration status did not
consider relationships with no age information, the table describes which types of relationships

93
and how many relationships of each type were excluded due to missing age information.
Participants reported up to five social network members and relationship connection type is
indicated by relationship type (row categories) and whether connections are intergenerational,
peer, or missing age information (column categories). The categories are not mutually exclusive,
such that participants can have relationship connection types across categories. For example, a
participant may have one intergenerational grandchild, one intergenerational child, a peer spouse,
and one missing age child. For those who reported having each relationship connection type
(e.g., intergenerational children, peer siblings, missing age other relatives), we calculated means,
standard deviations, and ranges. There are too many possible relationship connection
combinations given that participants can name up to five social network members who may all
be in one relationship connection category (e.g., 5 intergenerational children), or more likely,
categorized across relationship connection types in various ways. Even when the relationship
categories are condensed into family vs. non-family, there are too many combinations to derive
meaningful conclusions. In future analyses, it may be possible to impute some of the
intergenerational and peer statuses if we assume that missing age grandchildren are
intergenerational and missing age siblings are peers.
Regardless of relationship types, if we tried to include social network members missing
age information, there would also be more combinations of connection types, as displayed in
Appendix C. Not only does this categorization lead to small groupings, but also, it does not
provide meaningful information about age integration, as we do not know why age information
was excluded for certain social network members. There are several reasons why participants
may not have provided age information, but none can be confirmed. For example, did
participants not provide age information because they did not know it? If so, did they not know it

94
because they have memory problems, never knew the person’s age, or did not think age was an
important feature of connections? Perhaps participants simply did not mention an age because
they were not certain the exact number or did not want to assume their friend’s age. It is also
possible the interviewer did not ask age information for every social network member due to the
tedious data collection process or that the interviewer or participants thought the age information
from previous rounds did not need to be provided again in subsequent rounds.
Alternatively, the social network members listed might change over time, but the current
analyses only indicate whether presence/absence of peers at either wave or presence/absence of
intergenerational ties change. For example, if a participant listed their peer spouse and their
intergenerational daughter in the first wave, their age integration status would be age-integrated
at baseline. In the follow-up wave, the participant may list their peer spouse and a different
intergenerational tie, other than the previously named daughter. However, this cannot be deduced
from the current data set. As another example, that same participant may no longer have a peer
spouse at follow-up but may remain age-integrated by including another type of peer (i.e.,
sibling, other relative, child) in their social network. It should be noted that this limitation also
applies to social network changes that reflect lack of age information, rather than a true change
in one’s social network. Using the same participant in this example, age-integrated networks at
baseline may change to peer-only or intergenerational-only if the participant did not provide age
information for their peer spouse or intergenerational daughter at follow-up as well. The age
information from baseline could not be carried forward based on relationship type, because there
is no way to know that the spouse or daughter listed in follow-up rounds are the same individuals
from earlier rounds. Given the information provided by the missing age indicator for social
relationships listed, analyses could be conducted to compare the number and types of

95
relationships that were excluded from the analysis due to missing age information. Although
relationship types without age information could be used to designate intergenerational or peer
status, assumptions about non-relatives, other relatives, and even children and spouses cannot
reasonably indicate whether social ties are intergenerational or peer.
In some cases, if all age information was missing from social networks, participants were
excluded from the baseline (N = 1,245) and follow-up (N = 1,146) samples. Appendix D shows
the possible combinations of social network “age” patterns when including social network
members categorized as “Peer,” “Intergenerational,” and “Missing Age.” Again, though we do
not know why ages were missing, this information provides some insight on the “missing pieces”
of social networks in this data set, including how categories were combined to create the Age
Integration Status variable, whether participants had any missing age social network members or
not.
Conclusion
In conclusion, this dissertation is an important first step in social network literature
examining age integration within older adults’ social networks by considering not only
intergenerational ties, but also peer ties. These studies show that older adults have distinct
experiences of age integration within their social networks depending on whether they have
intergenerational or peer ties only, or both intergenerational and peer ties. The Social Convoy
Theory emphasizes that social experiences vary based on individual characteristics; the current
study contributes another social experience to consider. While there is a lot of information on
common social network features, like size, relationship quality, relationship type, and support
exchanges, we offer a new, but important, social network feature that is uniquely associated with
psychosocial outcomes and sociodemographic characteristics. There is more to learn about age

96
integration experiences in older adults’ social networks, and this dissertation provides the
infrastructure for future evaluations of age integration status. Future research may also consider
social network age integration from the perspectives of older adults, as their perception of being
age-integrated may be as important or more important than our objective identification that
certain older adults are age-integrated.

97
REFERENCES

Adams, K. B., Leibbrandt, S., & Moon, H. (2011). A critical review of the literature on social
and leisure activity and wellbeing in later life. Ageing and Society, 31(4), 683–712.
https://doi.org/10.1017/S0144686X10001091
Ajrouch, K. J., Antonucci, T. C., & Janevic, M. R. (2001). Social Networks Among Blacks and
Whites: The Interaction Between Race and Age. The Journals of Gerontology: Series B,
56(2), S112–S118. https://doi.org/10.1093/geronb/56.2.S112
Ajrouch, K. J., Blandon, A. Y., & Antonucci, T. C. (2005). Social networks among men and
women: The effects of age and socioeconomic status. Journals of Gerontology - Series B
Psychological Sciences and Social Sciences, 60(6), 311–317.
https://doi.org/10.1093/geronb/60.6.S311
Antonucci, T. C., Ajrouch, K. J., & Birditt, K. S. (2014). The Convoy Model: Explaining Social
Relations From a Multidisciplinary Perspective. The Gerontologist, 54(1), 82–92.
https://doi.org/10.1093/geront/gnt118
Antonucci, T. C., Ajrouch, K. J., & Webster, N. J. (2019). Convoys of social relations: Cohort
similarities and differences over 25 years. Psychology and Aging, 34(8), 1158–1169.
https://doi.org/10.1037/pag0000375
Antonucci, T. C., & Akiyama, H. (1987). An examination of sex differences in social support
among older men and women. Sex Roles, 17(11–12), 737–749.
https://doi.org/10.1007/BF00287685
Antonucci, T. C., Akiyama, H., & Takahashi, K. (2004). Attachment and close relationships
across the life span. Attachment and Human Development, 6(4), 353–370.
https://doi.org/10.1080/1461673042000303136

98
Ayalon, L., & Levkovich, I. (2019). A Systematic Review of Research on Social Networks of
Older Adults. The Gerontologist, 59(3), e164–e176. https://doi.org/10.1093/geront/gnx218
Baker, L. A., Cahalin, L. P., Gerst, K., & Burr, J. A. (2005). Productive activities and subjective
well-being among older adults: The influence of number of activities and time commitment.
Social Indicators Research, 73(3), 431–458. https://doi.org/10.1007/s11205-005-0805-6
Bektas, A., Schurman, S. H., Sen, R., & Ferrucci, L. (2018). Aging, inflammation and the
environment. Experimental Gerontology, 105, 10–18.
https://doi.org/10.1016/j.exger.2017.12.015
Berkman, L. F., & Syme, S. L. (1979). Social networks, host resistance, and mortality: a nine-
year follow-up study of Alameda County residents. American Journal of Epidemiology,
109(2), 186–204. https://doi.org/10.1093/oxfordjournals.aje.a112674
Berkman, L., Glass, T., Brissette, I., & Seeman, T. (2000). From social integration to health:
Durkheim in the new millennium. Social Science and Medicine, 51(6), 843–857.
https://doi.org/10.1016/S0277-9536(00)00065-4
Boehm, J. K., Chen, Y., Williams, D. R., Ryff, C. D., & Kubzansky, L. D. (2016). Subjective
well-being and cardiometabolic health: An 8-11year study of midlife adults. Journal of
Psychosomatic Research, 85, 1–8. https://doi.org/10.1016/j.jpsychores.2016.03.018
Canedo-García, A., García-Sánchez, J. N., & Pacheco-Sanz, D. I. (2017). A systematic review of
the effectiveness of intergenerational programs. Frontiers in Psychology, 8(OCT), 1–13.
https://doi.org/10.3389/fpsyg.2017.01882
Carlsson, G., & Karlsson, K. (1970). Age, Cohorts and the Generation of Generations. American
Sociological Review, 35, 710.

99
Carstensen, L. L. (1992). Social and emotional patterns in adulthood: Support for socioemotional
selectivity theory. Psychology and Aging, 7(3), 331–338. https://doi.org/10.1037//0882-
7974.7.3.331
Charles, S. T., & Carstensen, L. L. (2010). Social and emotional aging. Annual Review of
Psychology, 61, 383–409. https://doi.org/10.1146/annurev.psych.093008.100448
Cornwell, B. (2011). Independence through social networks: Bridging potential among older
women and men. Journals of Gerontology - Series B Psychological Sciences and Social
Sciences, 66 B(6), 782–794. https://doi.org/10.1093/geronb/gbr111
Cornwell, B. (2015). Social disadvantage and network turnover. Journals of Gerontology - Series
B Psychological Sciences and Social Sciences, 70(1), 132–142.
https://doi.org/10.1093/geronb/gbu078
Cornwell, B., Goldman, A., & Laumann, E. O. (2021). Homeostasis Revisited: Patterns of
Stability and Rebalancing in Older Adults’ Social Lives. The Journals of Gerontology.
Series B, Psychological Sciences and Social Sciences, 76(4), 778–789.
https://doi.org/10.1093/geronb/gbaa026
Cornwell, B., & Laumann, E. O. (2015). The health benefits of network growth: New evidence
from a national survey of older adults. Social Science and Medicine, 125, 94–106.
https://doi.org/10.1016/j.socscimed.2013.09.011
Cornwell, B., & Laumann, E. O. (2018). Structure by Death: Social Network Replenishment in
the Wake of Confidant Loss. 343–365. https://doi.org/10.1007/978-3-319-71544-5_16
Cornwell, B., Schumm, L. P., & Laumann, E. O. (2008). The social connectedness of older
adults: A national profile. American Sociological Review, 73(2), 185–203.
https://doi.org/10.1177/000312240807300201

100
Cornwell, B., Schumm, L. P., Laumann, E. O., & Graber, J. (2009). Social networks in the nshap
study: Rationale, measurement, and preliminary findings. Journals of Gerontology - Series
B Psychological Sciences and Social Sciences, 64(SUPPL.1).
https://doi.org/10.1093/geronb/gbp042
Cornwell, B., Schumm, L. P., Laumann, E. O., Kim, J., & Kim, Y. J. (2014). Assessment of
social network change in a national longitudinal survey. Journals of Gerontology - Series B
Psychological Sciences and Social Sciences, 69, S75–S82.
https://doi.org/10.1093/geronb/gbu037
Dannefer, D., & Feldman, K. (2017). Age Integration, Age Segregation, and Generation X: Life-
Course Perspectives. Generations, 41(3), 20–26.
Drury, L., Hutchison, P., & Abrams, D. (2016). Direct and extended intergenerational contact
and young people’s attitudes towards older adults. The British Journal of Social
Psychology. https://doi.org/10.1111/bjso.12146
Dykstra, P. A., & Fleischmann, M. (2016). Cross-Age friendships in 25 European countries.
Mens En Maatschappij: Tijdschrift Voor Sociale Wetenschappen, 91(2), 107–131.
https://doi.org/10.5117/MEM2016.2.DYKS
English, T., & Carstensen, L. L. (2014). Selective narrowing of social networks across adulthood
is associated with improved emotional experience in daily life. International Journal of
Behavioral Development, 38(2), 195–202. https://doi.org/10.1177/0165025413515404
Erikson, E. H. (Erik H. (1959). Identity and the life cycle; selected papers. International
Universities Press.

101
Fingerman, K. L., Huo, M., Charles, S. T., & Umberson, D. J. (2020). Variety Is the Spice of
Late Life: Social Integration and Daily Activity. Journals of Gerontology - Series B
Psychological Sciences and Social Sciences, 75(2), 377–388.
https://doi.org/10.1093/geronb/gbz007
Fiori, K. L., Antonucci, T. C., & Akiyama, H. (2008). Profiles of social relations among older
adults: a cross-cultural approach. Ageing and Society, 28(2), 203–231.
https://doi.org/10.1017/S0144686X07006472
Fiori, K. L., Antonucci, T. C., & Cortina, K. S. (2006). Social network typologies and mental
health among older adults. Journals of Gerontology - Series B Psychological Sciences and
Social Sciences. https://doi.org/10.1093/geronb/61.1.P25
Fiori, K. L., Smith, J., & Antonucci, T. C. (2007). Social network types among older adults: A
multidimensional approach. Journals of Gerontology - Series B Psychological Sciences and
Social Sciences. https://doi.org/10.1093/geronb/62.6.P322
Fried, L. P., Carlson, M. C., Freedman, M., Frick, K. D., Glass, T. A., Hill, J., Mcgill, S., Rebok,
G. W., Seeman, T., Tielsch, J., Wasik, B. A., & Zeger, S. (2004). A Social Model for Health
Promotion for an Aging Population: Initial Evidence on the Experience Corps Model. In
Journal of Urban Health: Bulletin of the New York Academy of Medicine (Vol. 81, Issue 1).
Gierveld, J. D. J., & Hagestad, G. O. (2006). Perspectives on the integration of older men and
women. Research on Aging, 28(6), 627–637. https://doi.org/10.1177/0164027506291871
Giraudeau, C., & Bailly, N. (2019). Intergenerational programs: What can school-age children
and older people expect from them? A systematic review. European Journal of Ageing,
16(3), 363–376. https://doi.org/10.1007/s10433-018-00497-4

102
Gruenewald, T. L., Karlamangla, A. S., Greendale, G. A., Singer, B. H., & Seeman, T. E. (2007).
Feelings of usefulness to others, disability, and mortality in older adults: The MacArthur
study of successful aging. Journals of Gerontology - Series B Psychological Sciences and
Social Sciences, 62(1). https://doi.org/10.1093/geronb/62.1.P28
Gruenewald, T. L., Liao, D. H., & Seeman, T. E. (2012). Contributing to others, contributing to
oneself: Perceptions of generativity and health in later life. Journals of Gerontology - Series
B Psychological Sciences and Social Sciences, 67 B(6), 660–665.
https://doi.org/10.1093/geronb/gbs034
Guo, M., Li, S., Liu, J., & Sun, F. (2015). Family Relations, Social Connections, and Mental
Health Among Latino and Asian Older Adults [Article]. Research on Aging, 37(2), 123–
147. https://doi.org/10.1177/0164027514523298
Hagestad, G. O., & Uhlenberg, P. (2005). The social separation of old and young: A root of
ageism. Journal of Social Issues, 61(2), 343–360. https://doi.org/10.1111/j.1540-
4560.2005.00409.x
Havighurst, R. J. (1961). Successful aging. The Gerontologist, 1, 8–13.
https://doi.org/10.1093/geront/1.1.8
Holt-Lunstad, J., Smith, T. B., & Layton, J. B. (2010). Social relationships and mortality risk: A
meta-analytic review. In PLoS Medicine (Vol. 7, Issue 7).
https://doi.org/10.1371/journal.pmed.1000316
House, J. S., Landis, K. R., & Umberson, D. (1988). Social Relationships. In Source: Science,
New Series (Vol. 241, Issue 4865).

103
Howell, R. T., Kern, M. L., & Lyubomirsky, S. (2007). Health benefits: Meta-analytically
determining the impact of well-being on objective health outcomes. Health Psychology
Review, 1(1), 83–136. https://doi.org/10.1080/17437190701492486
Huxhold, O., Fiori, K. L., Webster, N. J., & Antonucci, T. C. (2020). The strength of weaker ties:
An underexplored resource for maintaining emotional well-being in later life. Journals of
Gerontology - Series B Psychological Sciences and Social Sciences, 75(7), 1433–1442.
https://doi.org/10.1093/geronb/gbaa019
Huxhold, O., Fiori, K. L., & Windsor, T. D. (2013). The dynamic interplay of social network
characteristics, subjective well-being, and health: The costs and benefits of socio-emotional
selectivity. Psychology and Aging, 28(1), 3–16. https://doi.org/10.1037/a0030170
Jarrott, S. E., Weaver, R. H., Bowen, N. K., & Wang, N. (2018). Measuring dimensions of
intergenerational contact: factor analysis of the Queen’s University Scale. Aging and Mental
Health, 22(4), 568–573. https://doi.org/10.1080/13607863.2017.1280769
Jenkinson, C. E., Dickens, A. P., Jones, K., Thompson-Coon, J., Taylor, R. S., Rogers, M.,
Bambra, C. L., Lang, I., & Richards, S. H. (2013). Is volunteering a public health
intervention? A systematic review and meta-analysis of the health and survival of
volunteers. BMC Public Health, 13(1), 1–10. https://doi.org/10.1186/1471-2458-13-773
Joiner, R. J., Bergeman, C. S., & Wang, L. (2018). Affective experience across the adult
Lifespan: An accelerated longitudinal design. Psychology and Aging.
https://doi.org/10.1037/pag0000257

104
Kasper, J. D., & Freedman, V. A. (2014). Findings from the 1st round of the National Health and
Aging Trends Study (NHATS): introduction to a special issue. The Journals of
Gerontology. Series B, Psychological Sciences and Social Sciences.
https://doi.org/10.1093/geronb/gbu125
Kawachi, I., & Berkman, L. F. (2001). Social Ties and Mental Health. In Journal of Urban
Health: Bulletin of the New York Academy of Medicine (Vol. 78, Issue 3).
https://doi.org/10.1093/jurban/78.3.458
Lee, S., Koffer, R. E., Sprague, B. N., Charles, S. T., Ram, N., & Almeida, D. M. (2018).
Activity Diversity and Its Associations With Psychological Well-Being Across Adulthood.
The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 73(6),
985–995. https://doi.org/10.1093/geronb/gbw118
Levy, B. (2009). Stereotype Embodiment. Current Directions in Psychological Science, 18(6),
332–336. https://doi.org/10.1111/j.1467-8721.2009.01662.x
Levy, B. R., Zonderman, A. B., Slade, M. D., & Ferrucci, L. (2009). Age stereotypes held earlier
in life predict cardiovascular events in later life. Psychological Science, 20(3), 296–298.
https://doi.org/10.1111/j.1467-9280.2009.02298.x
Li, T., & Zhang, Y. (2015). Social network types and the health of older adults: Exploring
reciprocal associations. Social Science & Medicine, 130, 59–68.
https://doi.org/10.1016/j.socscimed.2015.02.007
Litwin, H. (2001). Social network type and morale in old age. Gerontologist, 41(4), 516–524.
https://doi.org/10.1093/geront/41.4.516

105
Litwin, Howard, Levinsky, M., & Schwartz, E. (2020). Network type, transition patterns and
well-being among older Europeans. European Journal of Ageing, 17(2), 241–250.
https://doi.org/10.1007/s10433-019-00545-7
Litwin, Howard, & Shiovitz-Ezra, S. (2011a). The association of background and network type
among older americans: Is “who you are” related to “who you are with”? Research on
Aging, 33(6), 735–759. https://doi.org/10.1177/0164027511409441
Litwin, Howard, & Shiovitz-Ezra, S. (2011b). Social network type and subjective well-being in a
national sample of older Americans. Gerontologist, 51(3), 379–388.
https://doi.org/10.1093/geront/gnq094
Luo, M., Ding, D., Bauman, A., Negin, J., & Phongsavan, P. (2020). Social engagement pattern,
health behaviors and subjective well-being of older adults: An international perspective
using WHO-SAGE survey data. BMC Public Health, 20(1), 1–10.
https://doi.org/10.1186/s12889-019-7841-7
McAdams, D. P., & de St. Aubin, E. (1992). A theory of generativity and its assessment through
self-report, behavioral acts, and narrative themes in autobiography. Journal of Personality
and Social Psychology, 62(6), 1003–1015. https://doi.org/10.1037/0022-3514.62.6.1003
Morack, J., Infurna, F. J., Ram, N., & Gerstorf, D. (2013). Trajectories and personality correlates
of change in perceptions of physical and mental health across adulthood and old age.
International Journal of Behavioral Development, 37(6), 475–484.
https://doi.org/10.1177/0165025413492605
Nguyen, Ann W., Chatters, L. M., Taylor, R. J., & Mouzon, D. M. (2016). Social Support from
Family and Friends and Subjective Well-Being of Older African Americans. Journal of
Happiness Studies, 17(3), 959–979. https://doi.org/10.1007/s10902-015-9626-8

106
Nguyen, Ann W. (2017). Variations in Social Network Type Membership Among Older African
Americans, Caribbean Blacks, and Non-Hispanic Whites. The Journals of Gerontology
Series B: Psychological Sciences and Social Sciences, 72(4), 716–726.
https://doi.org/10.1093/geronb/gbx016
Nieminen, T., Martelin, T., Koskinen, S., Simpura, J., Alanen, E., Härkänen, T., & Aromaa, A.
(2007). Measurement and Socio-Demographic Variation of Social Capital in a Large
Population-Based Survey. Social Indicators Research, 85, 405–423.
https://doi.org/10.1007/s11205-007-9102-x
Parisi, J. M., Kuo, J., Rebok, G. W., Xue, Q. L., Fried, L. P., Gruenewald, T. L., Huang, J.,
Seeman, T. E., Roth, D. L., Tanner, E. K., & Carlson, M. C. (2015). Increases in Lifestyle
Activities as a Result of Experience Corps?? Participation. Journal of Urban Health, 92(1),
55–66. https://doi.org/10.1007/s11524-014-9918-z
Pinquart, M. (2001). Gender differences in self-concept and psychological well-being in old age.
The Journal of Gerontology, 56(4), 195–213.
http://psychsocgerontology.oxfordjournals.org/content/56/4/P195.short
Pinquart, M. (2003). Loneliness in married, widowed, divorced, and never-married older adults
[Article]. Journal of Social and Personal Relationships, 20(1), 31–53.
https://doi.org/10.1177/0265407503020001186
Pinquart, M., & Sörensen, S. (2000). Influences of Socioeconomic Status, Social Network, and
Competence on Subjective Well-Being in Later Life: A Meta-Analysis Preparation for
Future Care Needs-general interest area View project How Effective Are Interventions With
Caregivers? View project. Psychology and Aging , 15(2), 187–224.
https://www.researchgate.net/publication/12438552

107
Pristavec, T. (2018). Social Participation in Later Years: The Role of Driving Mobility. Journals
of Gerontology - Series B Psychological Sciences and Social Sciences, 73(8), 1457–1469.
https://doi.org/10.1093/geronb/gbw057
Riley, M. W., & Riley, J. W. (1994). Age Integration and the Lives of Older People. The
Gerontologist.
Riley, M. W., & Riley Jr., J. W. (2000). Age integration: Conceptual and historical background.
The Gerontologist, 40, 266–270.
Rook, K. S. (1984). The negative side of social interaction: Impact on psychological well-being.
Journal of Personality and Social Psychology. https://doi.org/10.1037/0022-3514.46.5.1097
Rook, K. S., & Charles, S. T. (2017). Close social ties and health in later life: Strengths and
vulnerabilities. American Psychologist, 72(6), 567–577.
https://doi.org/10.1037/amp0000104
Roth, A. R. (2020). Informal caregiving and network turnover among older adults. Journals of
Gerontology - Series B Psychological Sciences and Social Sciences, 75(7), 1538–1547.
https://doi.org/10.1093/geronb/gby139
Rowe, J. W., & Kahn, R. L. (1997). Successful Aging1. The Gerontologist, 37(4), 433–440.
https://doi.org/10.1093/geront/37.4.433
Ryff, C. D., Radler, B. T., & Friedman, E. M. (2015). Persistent psychological well-being
predicts improved self-rated health over 9–10 years: Longitudinal evidence from MIDUS.
Health Psychology Open, 2(2), 2055102915601582.
https://doi.org/10.1177/2055102915601582

108
Saadeh, M., Welmer, A. K., Dekhtyar, S., Fratiglioni, L., & Calderón-Larrañaga, A. (2020). The
Role of Psychological and Social Well-being on Physical Function Trajectories in Older
Adults. The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences,
75(8), 1579–1585. https://doi.org/10.1093/gerona/glaa114
Sarkisian, N., & Gerstel, N. (2015). Does singlehood isolate or integrate? Examining the link
between marital status and ties to kin, friends, and neighbors. Journal of Social and
Personal Relationships, 33(3), 361–384. https://doi.org/10.1177/0265407515597564
Schafer, M. H. (2013). Structural Advantages of Good Health in Old Age: Investigating the
Health-Begets-Position Hypothesis With a Full Social Network. Research on Aging, 35(3),
348–370. https://doi.org/10.1177/0164027512441612
Schwartz, E., & Litwin, H. (2017). Are newly added and lost confidants in later life related to
subsequent mental health? International Psychogeriatrics, 29(12), 2047–2057.
https://doi.org/10.1017/S1041610217001338
Schwartz, E., & Litwin, H. (2018). Social network changes among older Europeans: the role of
gender. European Journal of Ageing, 15(4), 359–367. https://doi.org/10.1007/s10433-017-
0454-z
Schwartz, E., & Litwin, H. (2019). The Reciprocal Relationship Between Social Connectedness
and Mental Health Among Older European Adults: A SHARE-Based Analysis. The
Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 74(4), 694–
702. https://doi.org/10.1093/geronb/gbx131

109
Seeman, T. E., Lusignolo, T. M., Albert, M., & Berkman, L. (2001). Social relationships, social
support, and patterns of cognitive aging in healthy, high-functioning older adults:
MacArthur studies of successful aging. Health Psychology, 20(4), 243–255.
https://doi.org/10.1037/0278-6133.20.4.243
Sharifian, N., Manly, J. J., Brickman, A. M., & Zahodne, L. B. (2019). Social network
characteristics and cognitive functioning in ethnically diverse older adults: The role of
network size and composition. Neuropsychology, 33(7), 956–963.
https://doi.org/10.1037/neu0000564
Suanet, B., & Huxhold, O. (2020). Cohort Difference in Age-Related Trajectories in Network
Size in Old Age: Are Networks Expanding? Journals of Gerontology - Series B
Psychological Sciences and Social Sciences, 75(1), 137–147.
https://doi.org/10.1093/geronb/gbx166
Suanet, B., Van Tilburg, T. G., & Van Groenou, M. I. B. (2013). Nonkin in older adults’
personal networks: More important among later cohorts? Journals of Gerontology - Series B
Psychological Sciences and Social Sciences, 68(4), 633–643.
https://doi.org/10.1093/geronb/gbt043
Sun, H., & Schafer, M. H. (2019). Age integration in older Europeans’ non-kin core networks:
Does formal social participation play a role? European Journal of Ageing, 16(4), 455–472.
https://doi.org/10.1007/s10433-019-00507-z
Tang, F., Jang, H., Rauktis, M. B., Musa, D., & Beach, S. (2019). The race paradox in subjective
wellbeing among older Americans. Ageing and Society, 39(3), 568–589.
https://doi.org/10.1017/S0144686X17001064

110
Thoits, P. A. (2011). Mechanisms linking social ties and support to physical and mental health.
Journal of Health and Social Behavior, 52(2), 145–161.
https://doi.org/10.1177/0022146510395592
Tomioka, K., Kurumatani, N., & Hosoi, H. (2017). Self-rated health predicts decline in
instrumental activities of daily living among high-functioning community-dwelling older
people. Age and Ageing, 46(2), 265–270. https://doi.org/10.1093/ageing/afw164
Trudel-Fitzgerald, C., Millstein, R. A., Von Hippel, C., Howe, C. J., Tomasso, L. P., Wagner, G.
R., & Vanderweele, T. J. (2019). Psychological well-being as part of the public health
debate? Insight into dimensions, interventions, and policy. BMC Public Health, 19(1), 1–11.
https://doi.org/10.1186/s12889-019-8029-x
Uhlenberg, P. (2000). The Forum Essays on Age Integration Introduction: Why Study Age
Integration? In The Gerontologist (Vol. 40, Issue 3).
https://academic.oup.com/gerontologist/article-abstract/40/3/266/605362
Uhlenberg, P., & De Jong Gierveld, J. (2004). Age-segregation in later life: An examination of
personal networks. Ageing and Society, 24(1), 5–28.
https://doi.org/10.1017/S0144686X0300151X
Van Broese Groenou, M. I., & Van Tilburg, T. (2003). Network size and support in old age:
Differentials by socio-economic status in childhood and adulthood. Ageing and Society,
23(5), 625–645. https://doi.org/10.1017/S0144686X0300134X
van Tilburg, T. G., Suanet, B., & Carr, D. (2019). Unmarried Older People: Are They Socially
Better Off Today? The Journals of Gerontology Series B: Psychological Sciences and
Social Sciences, 74(8), 1463–1473. https://doi.org/10.1093/geronb/gby120

111
Varma, V. R., Carlson, M. C., Parisi, J. M., Tanner, E. K., McGill, S., Fried, L. P., Song, L. H.,
& Gruenewald, T. L. (2015). Experience corps Baltimore: Exploring the stressors and
rewards of high-intensity civic engagement. Gerontologist.
https://doi.org/10.1093/geront/gnu011
Vespa, J., Medina, L., & Armstrong, D. (2020). Demographic turning points for the United
States: Population projections for 2020 to 2060. Current Population Reports, P25-1144, 1–
15. https://census.gov/programs-surveys/popproj.html
Vozikaki, M., Linardakis, M., Micheli, K., & Philalithis, A. (2017). Activity participation and
well-being among European adults aged 65 years and older. Social Indicators Research,
131(2), 769–795. https://doi.org/10.1007/s11205-016-1256-y
Wrzus, C., Hänel, M., Wagner, J., & Neyer, F. J. (2013). Social network changes and life events
across the life span: A meta-analysis. Psychological Bulletin, 139(1), 53–80.
https://doi.org/10.1037/a0028601
Zautra, A. J., Reich, J. W., & Guarnaccia, C. A. (1990). Some Everyday Life Consequences of
Disability and Bereavement for Older Adults. Journal of Personality and Social
Psychology, 59(3), 550–561. https://doi.org/10.1037/0022-3514.59.3.550

112
Appendices
Appendix A: Row Percentages, Age Integration Status and Confounding Variables
Peer-Only Intergenerational-
Only
Age-Integrated
Marital Status
Married/Partnered 66.76 5.12 28.11
Divorced/Separated 39.64 47.95 12.41
Widowed 28.87 49.99 21.14
Never married 65.34 22.52 12.14
Parental Status
Non-Parent 98.44 1.56 0.00
Parent 52.98 21.50 25.52
Living Arrangement
Lives alone 30.61 49.57 19.82
Lives with spouse 67.45 4.80 27.75
Lives with others 66.00 5.67 28.33
Lives with spouse + others 38.01 44.81 17.19

113
Appendix B: Weighted Multinomial Logistic Regression, Well-Being and Social
Engagement on Age Integration Status

Peer-Only

N = 1,768
RRR [95% CI]
Intergenerational-
Only

N = 890
RRR [95% CI]
Independent Variable
Well-Being 1.01 [0.98, 1.04] 0.97 [0.93, 1.01]
Covariates
Age (67-101) 0.95 [0.93, 0.96]*** 1.06 [1.04, 1.07]***
Female* 0.73 [0.59, 0.89]** 1.60 [1.24, 2.05]***
Race/Ethnicity
Black, non-Hispanic 1.09 [0.81, 1.46] 2.02 [1.46, 2.80]***
Hispanic 0.82 [0.52, 1.30] 1.89 [1.13, 3.16]*
Other 0.61 [0.32, 1.15] 1.63 [0.79, 3.34]
More than HS Education 0.82 [0.67, 1.02] 1.15 [0.90, 1.49]
Self-Care Help 0.98 [0.74, 1.29] 0.57 [0.41, 0.78]***
Health Conditions 1.04 [0.96, 1.13] 1.01 [0.92, 1.10]
Peer-Only

N = 1,768
RRR [95% CI]
Intergenerational-
Only

N = 890
RRR [95% CI]
Independent Variable
Social Engagement 0.93 [0.86, 1.01] 0.81 [0.73, 0.89]***
Covariates
Age (67-101) 0.95 [0.93, 0.96]*** 1.06 [1.04, 1.07]***
Female* 0.74 [0.60, 0.91]** 1.72 [1.33, 2.21]***
Race/Ethnicity
Black, non-Hispanic 1.09 [0.81, 1.46] 1.94 [1.40, 2.69]***
Hispanic 0.80 [0.50, 1.27] 1.70 [1.01, 2.85]*
Other 0.60 [0.32, 1.12] 1.49 [0.72, 3.07]
More than HS Education 0.86 [0.69, 1.07] 1.28 [0.98, 1.66]
Self-Care Help 0.95 [0.72, 1.25] 0.55 [0.40, 0.75]***
Health Conditions 1.03 [0.95, 1.11] 1.00 [0.92, 1.09]

Note. Reference categories: age-integrated; male; White, non-Hispanic; high school education or
less; no self-care help. RRR = Relative Risk Ratio of having intergenerational-only or peer-only
compared to age-integrated social networks. * p < 0.05, ** p < 0.01, *** p < 0.001.

114

Appendix C: Relationship Types
Intergenerational

Peer Missing Age
% N M (SD) Min-
Max
% N M (SD) Min-
Max
% N M (SD) Min-
Max
Baseline
(N = 7,662)

Spouse 0.42 32 1.00 (0.00) 1-1 37.16 2,847 1.00 (0.00) 1-1 0.01 1 1.00 (0.00) 1-1
Parents - - - - - - - - - - - -
Siblings 0.04 3 1.00 (0.00) 1-1 0.52 40 1.13 (0.33) 1-2 14.83 1,136 1.27 (0.58) 1-5
Children 31.44 2,409 1.48 (0.76) 1-5 19.71 1,510 1.31 (0.59) 1-5 2.01 154 1.06 (0.24)
Grandchildren 0.56 43 1.09 (0.29) 1-2 - - - - 1.34 103 1.11 (0.37) 1-3
Other
Relatives
0.14 11 1.00 (0.00) 1-1 0.09 7 1.00 (0.00) 1-1 4.57 350 1.21 (0.48) 1-4
Non-Relatives - - - - - - - - 20.45 1,567 1.44 (0.73) 1-5
Follow-Up
(N = 5,553)

Spouse 0.32 18 1.00 (0.00) 1-1 34.29 1,904 1.00 (0.00) 1-1 - - - -
Parents 0.04 2 1.00 (0.00) 1-1 0.09 7 1.00 (0.00) 1-1 - - - -
Siblings 0.04 2 1.00 (0.00) 1-1 0.76 42 1.00 (0.00) 1-1 19.20 1,066 1.31 (0.61) 1-5
Children 34.32 1,906 1.50 (0.75) 1-5 22.67 1,259 1.30 (0.58) 1-4 2.54 141 1.11 (0.33) 1-3
Grandchildren 0.77 43 1.09 (0.37) 1-3 - - - - 2.22 123 1.15 (0.43) 1-3
Other
Relatives
0.16 9 1.22 (0.67) 1-3 0.09 5 1.20 (0.45) 1-2 6.77 376 1.21 (0.48) 1-3
Non-Relatives 0.04 2 1.00 (0.00) 1-1 0.11 6 1.00 (0.00) 1-1 30.38 1,687 1.57 (0.83) 1-5

115
Appendix D: Social Network Patterns, Intergenerational, Peer, and Missing Age
Connections
Baseline
N = 7,662
Follow-Up
N = 5,553
No Social Network Information 673 311
No Social Network Members 521 222
Intergenerational * 834 519
Peer + 1,909 1,087
Missing Age 1,245 1,146
Intergenerational and Peer ^ 885 626
Missing Age and Intergenerational* 411 455
Missing Age and Peer + 840 827
Missing Age, Peer, and IG ^ 344 360
Note. Categories combined into: * Intergenerational-Only, + Peer-Only, ^ Age-Integrated.

Abstract (if available)

Abstract As older adults live longer and have more opportunities to remain socially integrated with others, they can connect with individuals not only in their own generation, but also in younger generationsㅡthat is, older adults have more opportunities for age integration within their social networks. Prior research has compared those with and without intergenerational ties to understand age integration within older adults’ social networks; however, the lack of consideration for presence of same-aged peers suggests that age integration of social networks is only reflected by having intergenerational connections. We argue that a more accurate representation of age integration in social networks is having both same-aged peers and intergenerational connections because it considers the individuals’ social integration with several age groups, rather than just intergenerational connections. This dissertation fills a gap in the literature by defining age integration status of older adults’ social networks as age-integrated, intergenerational-only, and peer-only to identify whether, which, and how older adults benefit from age integration. ❧ The objective of this dissertation is to further our understanding of age integration in older adults’ social networks, including how age-integrated older adults vary from those with peer-only and intergenerational-only networks (Chapter 2), whether they derive unique psychosocial benefits due to their age-diverse social connections (Chapter 3), and how age integration status changes over time (Chapter 4). This dissertation consists of three studies which utilize secondary data from the National Health and Aging Trends Study (NHATS) to investigate sociodemographic, health, and psychosocial correlates of age integration status in a nationally representative sample of Medicare beneficiaries in the U.S.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Self-perceptions of Aging in the Context of Neighborhood and Their Interplay in Late-life Cognitive Health

PDF

Internet communication use, psychological functioning and social connectedness at older ages

PDF

Biomarkers of age-related health changes: associations with health outcomes and disparities

PDF

Estimating survival in the face of pain: evidence from the health and retirement study

PDF

The role of the modern aging network: measuring innovations of Area Agencies on Aging

PDF

Mechanisms of stress effects on learning and decision making in younger and older adults

PDF

The measurement, life course patterns, and outcomes of intergenerational ambivalence among parent-adult child dyads

PDF

Is stress exposure enough? Race/ethnic differences in the exposure and appraisal of chronic stressors among older adults

PDF

Intergenerational transmission of values and behaviors over the family life course

PDF

Implementation of peer providers in integrated health care settings

PDF

Residential care in Los Angeles: policy and planning for an aging population

PDF

Three essays on modifiable determinants of shingles: risk factors for shingles incidence and factors affecting timing of vaccine uptake

PDF

A biodemographic approach to understanding sociodemographic disparities in kidney functioning on three dimensions: individual, population, and cross-national

PDF

Examining the longitudinal influence of the physical and social environments on social isolation and cognitive health: contextualizing the role of technology

PDF

Social determinants of physiological health and mortality in China

PDF

Intergenerational support between grandparents and grandchildren in rural China and its effect on the psychological well-being of older adults

PDF

Statistical algorithms for examining gene and environmental influences on human aging

PDF

Cognitive health and self-rated memory in later life: linkages to race/ethnicity, multimorbidity, and survival

PDF

Synaptic transmission, nutrient sensors, and aging in Drosophila melanogaster

PDF

Computational approaches to identify genetic regulators of aging and late-life mortality

Asset Metadata

Creator Roman, Carly Jade (author)

Core Title Age integration in late life: sociodemographic & psychosocial correlates of intergenerational-only, peer-only, and age-integrated social networks

School Leonard Davis School of Gerontology

Degree Doctor of Philosophy

Degree Program Gerontology

Degree Conferral Date 2021-12

Publication Date 11/01/2021

Defense Date 05/11/2021

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

Tag age integration,intergenerational,OAI-PMH Harvest,social engagement,Social integration,social networks

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Zelinski, Elizabeth (committee chair), Ailshire, Jennifer (committee member), Beam, Christopher (committee member)

Creator Email carlyjader@gmail.com,carlyrom@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC16345213

Unique identifier UC16345213

Legacy Identifier etd-RomanCarly-10191

Document Type Dissertation

Format application/pdf (imt)

Rights Roman, Carly Jade

Type texts

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

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu