Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Dec 1.
Published in final edited form as: Adv Life Course Res. 2022 Aug 27;54:100502. doi: 10.1016/j.alcr.2022.100502

Linked Lives and Convoys of Social Relations

Noah J Webster 1, Toni C Antonucci 1,2, Kristine J Ajrouch 1,3
PMCID: PMC9849781  NIHMSID: NIHMS1844818  PMID: 36651615

Abstract

We consider linked lives through the Convoy Model of Social Relations to illustrate their complexity, consequences, and development across contexts. To illustrate how the Convoy Model lens provides a unique opportunity to examine the multidimensional and dynamic character of linked lives across time and space, we analyze twenty-three years of longitudinal data from the Social Relations Study (SRS). The SRS is a regionally representative Detroit-area sample (N=1,498) with three waves (1992; 2005; 2015) of data from community dwelling people age 13 to 93. We present three illustrative examples of linked lives: 1) the influence of earlier life social network characteristics (size and closeness) on later life health outcomes; 2) the influence of social position (race and education) on relationship quality with spouse/partner and child over time; and 3) the influence of transitioning from working to retirement on network structure (size and geographic proximity). Findings illustrate linked lives through multiple instances of social relationships and as influenced by various contexts. Further, the consequences of linked lives for mental health are consistent across the life course while influence on physical health is variable. The Convoy Model presents key concepts to situate the ways in which linked lives form and function at various levels and across multiple contexts to have far reaching effects on life outcomes.

Keywords: social convoys, social networks, relationship quality, mental and physical health

Two concepts are well established in the literature: linked lives and social relations. Each concept has made important contributions to theoretical understandings of life. Linked lives, a fundamental concept of the life course perspective refers to the ways in which an individual’s life trajectory is impacted by connections with others (Settersten, 2015). The concept acknowledges that peoples’ lives both affect and are affected by other people. In essence, linked lives are an instance of social relations. It is now understood that social relations are essentially universal and a primary source from which the self develops (Mead, 1934).

Recent theoretical and empirical work has demonstrated the complexity of social relations. Social relations are multidimensional, can be both positive and negative, are influenced by personal and situational factors, and have important consequences (Antonucci, 2001). Exploring how social relations and their consequences unfold over time and across contexts can advance understanding of how lives are linked (Settersten, 2015). Empirical data that include detailed measures of social relations overtime and across the life span however are lacking (Webster, Ajrouch, Wan, & Antonucci, 2017). How linked lives unfold across diverse contexts may be better elucidated using lifespan panel data that have detailed measures of social relations.

Acknowledgement of the importance of social relations and examination of how they and their influence on other life domains vary across critical contexts is not unique to the life course perspective. The social sciences more broadly have focused on these questions for some time. The life course perspective though, offers a unique lens through which to examine social relations. This involves integration of the linked lives concept with other fundamental concepts of the perspective (Carr, 2018), including temporality, a concept that is central to this paper.

Guided by the Convoy Model of Social Relations (Kahn & Antonucci, 1980) we use twenty-three years of longitudinal data from the U.S., regionally representative Social Relations Study (SRS) to examine the consequences of linked lives as well as contexts that are influential in their development. Specifically, we examine later life health consequences of earlier life social relations, and how social position (e.g., race and education) and situational change (e.g., role transitions such as work to retirement) exert influence on the development of the multiple dimensions of social relations over time. The Convoy Model lens provides unique opportunity to identify the multidimensional and dynamic character of linked lives across time and space.

Linked Lives and the Convoy Model of Social Relations

The Convoy Model of Social Relations (Kahn & Antonucci, 1980) conceptualizes social convoys as a unit, group or network of people with whom a person is linked. Convoys link people together, yet can vary in terms of emotional closeness. Convoys, and the people with whom a person is linked, are dynamic; that is, they change as both the individual and convoy members change. Convoy members are those most likely to either provide or receive support from the individual. The concept of a convoy is larger and more encompassing than a social network. Convoys are multidimensional, including, for example, structural characteristics of the network (e.g., size and geographic proximity of network members), but also support quality (e.g., satisfaction with or perceived adequacy of support). Further, support quality can be both positive and negative.

The Convoy Model recognition of social relations as dynamic (i.e., influenced by age-related processes) and multidimensional provides a unique framework in which to illustrate how linked lives form and function. In particular, the model integrates life span and life course perspectives (Antonucci, Fiori, Birditt, & Jackey, 2010). For instance, it specifies that the multidimensional components of social relations are shaped by contexts across time and place, fundamental principles of the life course perspective. Personal (e.g., age, race, educational attainment) and situational characteristics (e.g., role, event, socio-historical period) are fundamental contexts in which social relations, and therefore linked lives, should be examined. Finally, the Convoy Model argues that these personal and situational characteristics, social networks and relationship quality, all influence the individual in important ways.

The Convoy Model, when originally developed, focused on the individual, i.e., how a person perceived and experienced their relationship with others. While appropriate in the early stages of model development, only limited attention was given to how lives are linked. As the model was further developed, the bidirectional links between an individual and the people with whom they are linked were acknowledged. Specifically, individual characteristics (e.g., age) and the situations in which an individual is embedded (e.g., work) can influence the type, quantity, and quality of one’s relationships. These relationships, in turn, impact the individual in profound ways, such as short- and long-term health outcomes.

The Convoy Model provides a theoretical framework with which to understand the consequences and antecedents of linked lives in context. Guided by the model we seek to further understand the complexities of linked lives as they unfold over time. We demonstrate this linkage with three illustrative examples. First, we examine the long-term impact of earlier life social relations on later life mental and physical health and consider linked lives through an examination of social network characteristics. Here, we highlight the consequences of linked lives on the individual, and in doing so also examine how this process unfolds both over time and differently across the life span (i.e., differently across age contexts). In the second and third examples, we highlight how the contexts in which an individual is embedded influence how they are linked with others. The second example examines the influence of social position, as indicated by race and education, on quality of social relations (both positive and negative) over time. In the third example, we demonstrate the influence of situational change, i.e. the transition from work to retirement, on the network characteristics of size and geographic proximity over time. In the paragraphs below, we briefly review relevant literature for each of the three examples.

Consequences of linked lives on later life health

Social relations are widely recognized to influence health. Less is known about how social relations experienced earlier in life influence later life outcomes. The size and closeness of social networks indicate key elements of linked lives across the life course. There may be value in identifying at which points in the life course or in which age contexts these dimensions of social networks have the most impact on later life outcomes.

Network size and health:

Network size refers to the number of individuals one defines as close and important. It is theoretically critical in the study of social determinants of health because it indicates the extent to which an individual is linked to others (Berkman, Glass, Brissette, & Seeman, 2000). This is a key element of integration, and hence represents a fundamental indicator of linked lives. Social network size predicts a myriad of health outcomes, including mortality risk (e.g., Holt-Lunstad, Smith, & Layton, 2010) and depression (Fiori, Antonucci, & Akiyama, 2008). Larger networks imply the availability of more resources, yet also may signify opportunities for greater relationship strain (Antonucci, Akiyama, & Lansford, 1998). The effects of social network size on health may depend on age. For example, the benefits of larger networks in mid-life may be stronger on health outcomes than the very same characteristics experienced in young adulthood. Lachman & Agrigoroaei (2010) found social relations to benefit health in the oldest age, while Carstensen, Isaacowitz & Charles, 1999) have argued that as people age, the size of their social networks get smaller, presumably because they retain only those with whom they have meaningful relationships. Network size in mid-life, therefore, may be an especially significant predictor of later health, while the same characteristics among younger adults may not be as influential on mid-life health outcomes. How lives are linked and their importance may vary over the life course.

Relationship closeness and health:

Another key instance of linked lives is the matter of relationship closeness. The ability to distinguish the closest relationships from the less close (Antonucci & Akiyama, 1987), or strong versus weak ties (e.g., Granovetter, 1973), advances understanding as to whether social ties overall are important for health, or whether the closeness and importance of a tie matters. For example, according to Portes (1998; p. 5,) weaker ties ‘can be sources of new knowledge and resources.’ This may explain why weaker ties can facilitate better emotional well-being (Huxhold, Fiori, Webster, & Antonucci, 2020) and cognitive health (Pan & Chee, 2020). In contrast, closer ties tend to be those that are relied upon for emotional support as well as sick care and have been linked to the receipt of positive support and better functional health later in life (Webster, Antonucci, Ajrouch, & Abdulrahim, 2015). In terms of closeness, having larger numbers of close ties in mid and later life may be more strongly related to better later life health since those immediate resources may be needed to address emerging and ongoing health problems. In contrast, larger numbers of less close ties earlier in life represents social integration which can connect people to a wide range of resources that may be important for health in mid-life. Examining linked lives through analysis of whether social network change over time influences health status differentially by age provides an opportunity to identify the far-reaching effects of a specific aspect of linked lives at particular points in the life course.

Linked lives in relation to social position

Race and education are two social positions that represent enduring contexts in which linked lives develop, grow and take shape. These characteristics represent key indicators of inequality, shaping the challenges and resources available to maximize the quality of relationships with close others. While race and education are distinct measures of social position (e.g., race as ascribed and education achieved) and tap into unique forms of disadvantage (e.g., discrimination and socio-economic status) there is overlap. For example, race is often viewed as a proxy for many intertwined social processes including access to educational opportunities. Yet, it is important to examine each’s unique effect on how lives are linked.

Attention to the function of linked lives includes the extent to which the quality of relationships is positive and/or negative. According to Lin’s theory of social capital (2001), resources are also embedded within and accessed through linkages with other people. Race and education play a role in this process by shaping social network composition through homophily (i.e., people are linked with others who share similar characteristics). Considerable racial and educational homophily have been found within social networks (McPherson & Smith-Lovin, 2001). This suggests, for example, that someone who has higher levels of education may have network members with more education who are, therefore, able to provide more instrumental and emotional support (i.e., positive relationship quality).

Two relationship types that arguably illustrate the influence of linked lives more so than others are those with: 1) spouse/partner and 2) child. As a result, relationship quality with spouse/partner and child is of particular significance. Marriage patterns are known to vary by race. For example, a nationally representative study of kinship networks in the U.S. showed that White individuals are more likely than Black individuals to have a spouse in all age groups (Daw, Verdery, & Margolis, 2016). Factors that contribute to a lower likelihood of marriage among Black individuals, such as stress from institutional discrimination and accompanying financial hardships, are also theorized to place greater strain on the relationships of Black individuals relative to White individuals (Broman, 1993) when they do marry. Conflictual elements of the marital relationship tend to be higher among Black compared with White or Hispanic couples (Bulanda & Brown, 2007) and Black Caribbean couples (Bryant, Taylor, Lincoln, Chatters, & Jackson, 2008). Older age, however, is associated with greater marital quality among Black individuals (Bryant et al., 2008). Yet, it is unclear as to whether race differences in spousal relationship quality persist over time.

Education level is an indicator of human capital, which may ensure positive relationships with a child over time (Birditt, Hartnett, Fingerman, Zarit, & Antonucci, 2015). Education may also signify different styles of interacting between parent and child (Chen & Berdan, 2006; Lareau, 2002), which then are experienced as positive or negative. Though studies show that relationship quality between parent and child is influenced by gender and poverty level of parents (Belle, 1983; Fingerman, Huo, & Birditt, 2020), education level is often only considered as a covariate. The question remains as to whether education level influences relationship quality between parents and children over time. Examining the extent to which key contexts of stratification influence relationship quality across various relationship types over time will illuminate the diversity of linked lives.

Linked lives in relation to situational change

The life course perspective emphasizes the importance of transitions in shaping life outcomes (George, 1993). The notion that the work to retirement transition leads to network changes highlights how various situational contexts impact linked lives. A number of studies have found that life events can impact changes in social networks and network size in particular (Wrzus, Hael, Wagner, & Neyer, 2012). The literature focused on retirement linked changes in network size offers conflicting perspectives. Some have argued that retirement brings about social loss due to a reduction of ties with co-workers (e.g., Howard et al., 1982), while others argue that there should be continuity in social resources following retirement (Chappell & Havens, 1985) or even growth due to increased time for social activities (Palmore et al., 1979). A number of earlier studies tested these competing hypotheses (e.g., see van Tilburg, 2003). Most found no impact of retirement on network size (Fletcher, 2014; van Tilburg, 2003). Van Tilburg (2003), however, did find that retirement is associated with a reduction in co-worker ties, which suggests that network losses are being compensated for by new ties. Similarly, Mor-Barak and colleagues (1992) found that working in later life was associated with a larger number of friends. Further, studies have found that recent cohorts of retirees are more likely to include co-workers in their networks post-retirement (Cozijnsen, Stevens, & van Tilburg, 2010). Such findings support the perspective that retirement may be moving from a time of social continuity to growth. Hence, it may be that while in the past there was minimal change in the number of people in one’s network resulting from a transition to retirement, more recent cohorts may experience increases as they retain relationships with co-workers, but also develop new ties.

The majority of studies on this topic have focused on social network composition and size or the number of specific relationships (e.g., friends, co-workers). Both dimensions of network structure have important implications for linked lives, but one area that receives limited attention is geographic proximity. Retirement is often accompanied by relocation (Litwak & Longino, 1987), and network members may remain constant, but from afar. We argue that examination of the impact of retirement on additional measures of social networks are needed. Therefore, we examine both the impact of retirement on social network size and the impact on changes in network geographic proximity. Further, our study uses longitudinal data to better specify the contexts in which linked lives may change by comparing people who retire with those continuously working. We leverage empirical methods (e.g., mixed effects models with time varying and invariant predictors) that have been used to examine other distal influences, e.g., child poverty and parental divorce (e.g., Burnett & Farkas, 2009).

Present study

We investigate linked lives over time by considering patterns of network structure and relationship quality longitudinally including antecedents and consequences. Using the Convoy Model as the lens through which to identify key concepts for better instantiating linked lives, we present three illustrative examples to address the following research questions:

  1. Do early life experiences of social networks influence later life mental and physical health?

  2. Does social position, as indicated by race and education, influence changes in relationship quality with a spouse/partner and child?

  3. Does social network structure (size and geographic proximity) change across the work to retirement transition?

METHODS

Data are from the longitudinal Social Relations Study (SRS) which began in 1992 (Antonucci & Akiyama, 1994) and were collected via face-to-face interviews. The original sample was drawn from the tri-county Detroit metropolitan area. A two-stage area probability sample design was used to select a probability sample of housing units. Field interviewers visited each selected housing unit, completed a household roster, and selected a random household member age 13 and older to interview (N=1,498). Wave 1 had a 73% response rate. Wave 2, completed in 2005, using both telephone and face-to-face interviewing, included 1,076 of the original participants with 320 identified as deceased (response rate = 78%). Wave 3, completed in 2015, using telephone interviews included 720 of the original participants (response rate = 73%). Specific sub-samples from the SRS were selected to examine each of the three research questions.

Measures

1). Consequences of linked lives on later life health

Health.

To measure mental and physical health we used two measures from Waves 1–3: 1) Mental health was measured as depressive symptoms using the Center for Epidemiologic Studies Depression (CES-D) 20-item scale (Radloff, 1977). Items were asked on a 4-point scale (0=rarely/none of the time; 3=most of the time), which were summed; 2) Physical health was measured as self-rated health using a single item, “How would you rate your health at the present time?” on a 5-point scale (1=poor; 5=excellent).

Total network size and closeness

Total network size and closeness were assessed at Wave 1 using the Hierarchical Mapping Technique (Antonucci, 1986). Respondents were shown a diagram with a set of three concentric circles and a smaller circle in the center with the word “you”. They were asked to name people who were close, closer and closest. The total number of people in all three circles was then summed. Total network size was the number of people nominated in all three circles. Closeness was measured using three variables including the number of people nominated in each the inner (closest), middle (closer), and outer (close) circles.

Covariates.

We controlled for age at Wave 1, categorized as 13–24; 25–39; 40–59; and 60–86 year). Smaller age categories were created among the younger ages given the significantly different experiences that are normative for these earlier ages as well as to allow for more finite exploration of the impact of earlier life social relations on health. We controlled for gender (0=male, 1=female) and race (0=non-White; 1=White) from Wave 1, years of education completed from Wave 2 (centered at 12 years) to allow for the youngest age group to obtain their likely highest level of education, which for some may have still been increasing from Wave 1 to 2, and marital status in all three waves (0=not married/living with partner, which includes those never married, separated, divorced or widowed; 1=married/living with partner).

2). Linked lives in relation to social position

Relationship quality:

Linked lives in this example was conceptualized as relationship quality, both positive and negative with a spouse/partner and child relied on most, collected at all three waves. Positive relationship quality was measured separately for both relationships with 5-items including: my (relationship type) supports me; I enjoy being with my (relationship type); I can share my very private feelings and concerns with my (relationship type); my (relationship type) encourages me; and my (relationship type) believes in me. Each item was asked on a 5-point scale (1=disagree; 5=agree) and responses were averaged. Negative relationship quality was similarly measured using 2-items (e.g., my child gets on my nerves; makes too many demands on me).

Social position

Social position was measured using two variables collected at Wave 1: 1) race (0=Black; 1=White); and 2) education (years of education completed, centered at 12 years).

Covariates

Covariates included: age (year of birth centered at age 40) and gender (0=male; 1=female) at Wave 1, and marital status (0=not married/living with partner; 1=married/living with partner) and employment status (0=not employed; 1=working full or part-time) were collected at all three waves.

3). Linked lives in relation to situational change

Network Characteristics:

Two measures were assessed at all three waves to indicate linked lives: 1) Network size was measured using the Hierarchical Mapping Technique described above. 2) Geographic proximity was measured as percentage of respondents’ network who were geographically proximate (i.e., lives within an hour’s drive).

Work status

Work status was measured using two dummy variables created from employment status at all three waves to document the transition to retirement. The reference category was respondents who worked at all three waves. The dummy variables correspond to a) worked for 1 or 2 waves then reported being retired at Waves 2 or 3 and b) other work trajectories.

Covariates

Covariates included: age (year of birth centered at 40), gender (0=male; 1=female), race (0=non-White; 1=White), and years of education completed (centered at 12 years) from Wave 1, and marital status (0=not married/living with partner; 1=married/living with partner) collected at all three waves.

Analysis Strategy

1). Consequences of linked lives on later life health:

We conducted eight multi-level models, four for each health outcome. Multi-level models were conducted to account for the non-independent data structure (repeated measures nested within respondents). In each model, we included random effects for intercept and linear time (i.e., survey year: 0, 13, 23). Each respondent is assumed to have a unique intercept and linear slope for time. One set of models included total network size as a predictor, and the second set included closeness (i.e., inner, middle and outer circle sizes). Each model also included survey year and survey year by social network interactions. A second model was then tested for each by adding age × survey year × social network interactions. For mental health, we tested network interactions with both linear and quadratic survey year given the curvilinear change in this outcome. Significant interactions between social network characteristics and survey year were explored using an online tool to compute simple slopes for multi-level models (Preacher, Curran, & Bauer, 2006).

2). Linked lives in relation to social position:

We conducted eight multi-level models using the same procedures as described above, four each predicting positive and negative relationship quality with mother and child. Included in each model were our two measures of social position (race and education), survey year, and a social position × survey year interaction. Separate models were conducted to test the race × survey year and education × survey year interactions.

3). Linked lives in relation to situational change:

We conducted two multi-level models, using the same procedures described above, one for each outcome (network size and geographic proximity) with survey year, work-retirement status, and a survey year × work-retirement status interaction included as independent variables.

Missing data were handled in all analyses using listwise deletion by survey year, i.e., if a respondent had missing data at a specific wave only that wave of data was excluded from the analysis. Additionally, respondents were required to have complete data for at least two waves to be included in each analysis. For the later life health analysis, this could be W1 and 2 or W1 and 3, but for the other two analyses, respondents needed at least W1 and 3 data.

RESULTS

Consequences of linked lives on later life health

Mental health.

We first examined the association between Wave 1 social relations characteristics and mental health. Individuals who reported having larger social networks in Wave 1 reported better mental health (i.e., fewer depressive symptoms) (Table 2). We determine that the effect of Wave 1 network size on mental health is present and generally consistent in all three waves because there were no statistically significant interactions between network size and survey year. The analysis by circle placement (closeness) revealed that the effects were significant for the middle but not inner or outer circles. Individuals who reported having a larger number of middle circle members reported better mental health in all three waves.

Table 2.

Earlier life social network size effects on later life physical and mental health

Mental health (depressive symptoms)
 Physical (self-rated) health
Network size Inner, middle, outer circle sizes Network size Inner, middle, outer circle sizes
Model 1a Model 2b Model 1a Model 2b Model 1a Model 2b Model 1a Model 2b
b (SE) b (SE) b (SE) b (SE) b (SE) b (SE) b (SE) b (SE)
Intercept 15.41*** (0.98) 15.47*** (2.04) 15.21*** (0.99) 16.02*** (2.08) 4.00*** (0.10) 3.98*** (0.20) 4.00*** (0.10) 3.96*** (0.20)
Years of education (W1, centered at 12) −0.37*** (0.09) −0.37*** (0.09) −0.37*** (0.09) −0.37*** (0.09) 0.05*** (0.01) 0.05*** (0.01) 0.05*** (0.01) 0.05*** (0.01)
White −1.09* (0.54) −1.06 (0.54) −0.96 (0.55) −0.95 (0.55) 0.21*** (0.06) 0.21*** (0.06) 0.21*** (0.06) 0.21*** (0.06)
Female 0.77 (0.48) 0.86 (0.48) 0.67 (0.48) 0.81 (0.48) −0.02 (0.05) −0.02 (0.05) −0.02 (0.05) −0.02 (0.05)
Married or lives with partner −2.08*** (0.41) −1.81*** (0.42) −2.08*** (0.41) −1.79*** (0.42) 0.09* (0.04) 0.10* (0.04) 0.09* (0.04) 0.10* (0.04)
Survey year −0.05 (0.04) −0.12 (0.12) −0.05 (0.04) −0.16 (0.12) −0.02*** (0.00) −0.03** (0.01) −0.02*** (0.00) −0.03* (0.01)
Age (W1) 25–39c −1.70* (0.77) −0.70 (2.29) −1.48 (0.77) −1.54 (2.32) −0.18* (0.08) −0.08 (0.22) −0.19* (0.08) −0.07 (0.22)
Age (W1) 40–59c −1.97* (0.78) −2.96 (2.33) −1.86* (0.78) −3.49 (2.36) −0.33*** (0.08) −0.49* (0.22) −0.33*** (0.08) −0.48* (0.23)
Age (W1) 60–89c −2.66** (0.87) −4.33 (2.49) −2.55** (0.87) −5.03* (2.52) −0.52*** (0.09) −0.35 (0.24) −0.52*** (0.09) −0.35 (0.24)
Network size (W1) −0.16** (0.06) −0.12 (0.19) 0.00 (0.01) 0.01 (0.02)
Inner circle size (W1) −0.09 (0.09) −0.38 (0.32) 0.00 (0.01) 0.01 (0.03)
Middle circle size (W1) −0.33*** (0.09) −0.19 (0.34) 0.01 (0.01) 0.02 (0.03)
Outer circle size (W1) 0.07 (0.13) 0.26 (0.37) −0.02 (0.01) −0.01 (0.04)
Model 2—significant 2-way interactions
Survey year * Age 40–59 0.03* (0.01) 0.03* (0.01)
Survey year * Age 60–89 0.32* (0.16) 0.38* (0.16)
Model 2 – significant 3-way interactions
Network size * Survey year * Age 40–59 −0.003* (0.00)
Outer circle size * Survey year * Age 40–59 −0.01** (0.00)
SD (Intercept) 6.15 6.08 6.1 6.02 0.56 0.56 0.56 0.55
SD (Survey year) 0.21 0.19 0.21 0.19 0.02 0.02 0.02 0.02
N-Level 1 2,294 2,294 2,291 2,291 2,354 2,354 2,351 2,351
N-Level 2 890 890 889 889 907 907 906 906
AIC 16,310 16,291 16,292 16,291 6,054 6,049 6,050 6,053
LRT Chi-squared (M0, M1)d 104.6*** (6, 15) 37.7*** (15, 24) 111.3*** (6, 19) 42.54** (19, 40) 110.9*** (6, 15) 23** (15, 24) 115.3*** (6, 19) 38.95** (19, 40)
Open in a new tab

Notes:

*

p<0.05;

**

p<0.01;

***

p<0.001;

a

Model 1 included interactions for network size (and each specific circle size) × survey year, and none were statistically significant (p<.05) indicating that the main effect of network size is generally consistent in all three waves;

b

Model 2 tested three-way interactions between network size (and each specific circle size) × survey year × age group. All lower order (two-way) interactions were included in model 2 and only statistically significant (p<.05) three-way interactions are presented;

c

Age 13–24 is the reference group;

d

For model 1, likelihood ratio test chi-squared statistics calculated by comparing model 1 to an unconditional growth curve model with no covariates. For Model 2, likelihood ratio test chi-squared statistics calculated by comparing model 2 to model 1.

In the full models, there were no significant interactions between survey year and social network size, including all three circle sizes, indicating that change in mental health over time was not influenced by Wave 1 network size or closeness. There were also no significant interactions between age group, survey year and social network size or closeness indicating that the findings were consistent across age groups.

Physical health.

Models predicting physical health as a function of social network size revealed no significant main effects of network size or closeness on health (Table 2). There were also no significant interactions between social network size or closeness and survey year indicating that physical (i.e., self-rated) health did not show different patterns over time as a function of social network size in Wave 1.

There was, however, an interaction between network size, age, and linear survey year indicating that the association between total network size and physical health over time varied by age group. As seen in Figure 1, the interaction revealed that there was no change over time in physical health among 13–24 and 60–86 year-olds with larger networks, that is, 16 network members (80th quantile). All other comparison groups, including all age groups with network sizes of 6 (20th quantile), are predicted to decline in physical health over time, though to varying degrees. With respect to closeness, there was also an interaction between outer circle size, age group, and linear survey year indicating that the association between outer circle size and self-rated health over time varied by age group. As seen in Figure 2, findings for the outer circle were substantively similar to those for total network size, though varied in statistical significance when comparing respondents with no outer circle ties (20th quantile) to those with 4 (80th quantile). As with total network size, 13–24 and 60–86-year-olds with larger outer circles had positive, non-significant slopes while all remaining groups had negative slopes. In this case, only three of the remaining six groups had significant negative slopes: respondents age 25–39 with small outer circles, 40–59 with small outer circles, and 40–59 with large outer circles.

Figure 1.

Figure 1.

Open in a new tab

Physical health over time by wave 1 network size and age

Figure 2.

Figure 2.

Open in a new tab

Physical health over time by wave 1 outer circle size and age

Linked lives in relation to social position

We examined how race and education, both of which are associated with inequalities over the life course, predict change in linked lives through an examination of relationship quality with two family relationships: spouse/partner and child. We examined those who are aged 40+ at Wave 1, and find that linked lives change in complex ways. Participants were on average 51.5 years old (SD=8.5) at Wave 1, ranging in age from 40 to 77, and 61.3% were female. Participants reported on average 13.8 years of education (SD=2.4). Sixty-six percent of the sample was married and 80% were White.

As seen in Table 3 and Figure 3, positive relationship quality with spouse/partner increased slightly over time for Black but not White respondents. Spouse/partner negative relationship quality was stable over time in the linear model and did not vary by race or education. Alternative models (not shown in the table) indicate a curvilinear relationship over time, with negative relationship quality increasing from Wave 1 to 2, then decreasing from Wave 2 to 3. Positive relationship quality with child was stable over time with no moderating effect of race or education. In contrast, negative relationship quality decreased over time and there was a significant survey year × education interaction. Respondents with higher education levels showed a greater decrease over time in child negative relationship quality compared to those with lower education levels (see Figure 4).

Table 3.

Race and education effects on relationship quality over time

Spouse/Partner Relationship Quality
 Child Relationship Quality
Positive Negative Positive Negative
Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 Model 1 Model 2
b (SE) b (SE) b (SE) b (SE) b (SE) b (SE) b (SE) b (SE)
Intercept 4.63*** (0.17) 4.85*** (0.16) 2.21*** (0.35) 2.00*** (0.32) 4.33*** (0.12) 4.33*** (0.11) 3.12*** (0.24) 3.18*** (0.22)
Age (W1, centered at 40) 0.00 (0.00) 0.00 (0.00) −0.01 (0.01) −0.01 (0.01) 0.01** (0.00) 0.01** (0.00) −0.04*** (0.01) −0.04*** (0.01)
Years of education (W1, centered at 12) 0.02 (0.01) 0.03 (0.02) −0.10*** (0.03) −0.11** (0.03) −0.01 (0.01) −0.03 (0.02) −0.001 −0.01 (0.03)
White 0.26* (0.12) 0.01 (0.09) −0.13 (0.24) 0.13 (0.18) −0.12 (0.09) −0.08 (0.08) 0.28 (0.19) 0.11 (0.14)
Female −0.22*** (0.06) −0.22*** (0.06) 0.34** (0.13) 0.34** (0.13) 0.18** (0.06) 0.18** (0.06) −0.08 (0.12) −0.08 (0.12)
Married or lives with partner −0.13 (0.12) −0.13 (0.12) 0.42 (0.25) 0.42 (0.25) 0.15** (0.05) 0.14** (0.05) −0.35** (0.11) −0.32** (0.11)
Working 0.00 (0.05) −0.01 (0.05) 0.12 (0.10) 0.13 (0.10) −0.02 (0.04) −0.01 (0.04) −0.08 (0.10) −0.10 (0.10)
Survey year 0.02** (0.01) 0.00 (0.00) −0.02 (0.01) 0.00 (0.01) 0.00 (0.00) 0.00 (0.00) −0.02 (0.01) −0.02*** (0.01)
Survey year * White −0.02** (0.01) 0.02 (0.01) 0.00 (0.00) −0.01 (0.01)
Survey year * Education 0.00 (0.00) 0.00 (0.00) 0.00 (0.00) −0.003* (0.00)
SD (Intercept) 0.42 0.44 0.86 0.87 0.45 0.45 0.84 0.84
SD (Survey year) 0.00 0.01 0.03 0.03 0.01 0.01 0.04 0.04
N-Level 1 555 555 555 555 709 709 708 708
N-Level 2 228 228 228 228 257 257 257 257
AIC 828 838 1,634 1,636 980 977 2,150 2,147
LRT Chi-squared (M0, M1)a 9.99** (12, 13) 0.37 (12, 13) 2.82 (12, 13) 0.6 (12, 13) 0.54 (12, 13) 2.85 (12, 13) 1.51 (12, 13) 3.95* (12, 13)
Open in a new tab

Notes:

*

p<0.05;

**

p<0.01;

***

p<0.001;

a

Likelihood ratio test chi-squared statistics calculated by comparing the models to models without an interaction term.

Figure 3.

Figure 3.

Open in a new tab

Positive relationship quality with spouse/partner over time by race

Figure 4.

Figure 4.

Open in a new tab

Negative relationship quality with child over time by education.

Linked lives in relation to situational change

We examined differences in social networks between those who worked and then retired and those who worked continuously over 23 years. Table 1 reports the descriptive statistics for this sample. Participants were on average 40.2 years old (SD=13.1) at Wave 1, and ranged from ages 13 to 77, and 63.8% were female. Participants reported on average 13.5 years of education (SD=2.3). Sixty-four percent of the sample was married at Wave 1 and 75% were White.

Table 1.

SRS sample characteristics by example

Example 1
later life health (N=2,358)		 Example 2
social position (N=837)		 Example 3
situational change (N=1,797)
M (SD) M (SD) M (SD)
Age (W1) [13,77] 40.16 (13.14)
Age (W1) [40,77] 51.47 (8.45)
 13–24 11.8%
 25–39 36.5%
 40–59 34.1%
 60–86 17.6%
Education (W1) [4,17] 13.81 (2.37) 13.5 (2.29)
Education (W2) [1,17] 13.47 (2.64)
Race (White) 73.3% 80.3% 75.2%
Gender (Female) 61.7% 61.3% 63.8%
Married or lives with partner
 Wave 1 59.0% 72.0% 64.6%
 Wave 2 65.3% 66.3% 70.6%
 Wave 3 64.6% 58.6% 64.4%
Working full- or part-time
 Wave 1 70.3%
 Wave 2 44.8%
 Wave 3 21.1%
Work trajectory
 Worked continuously 39.1%
 Worked then retired 31.2%
 Other work trajectory 29.7%
Network size [0,20] 10.49 (5.25)
 Wave 1 11.43 (5.41) 10.84 (5.38)
 Wave 2 11.71 (5.34) 11.92 (5.25)
 Wave 3 10.33 (5.33) 10.80 (5.3)
Inner circle size [0,20] 4.27 (3.32)
Middle circle size [0,20] 4.15 (3.29)
Outer circle size [0,13] 2.06 (2.46)
Network Geographic Proximity (% within 1 hr drive)
 Wave 1 77.3%
 Wave 2 68.4%
 Wave 3 71.2%
Positive relationship quality, spouse/partner [1.4,5]
 Wave 1 4.69 (0.62)
 Wave 2 4.71 (0.51)
 Wave 3 4.75 (0.51)
Negative relationship quality, spouse/partner [1,5]
 Wave 1 2.5 (1.21)
 Wave 2 2.56 (1.14)
 Wave 3 2.39 (1.09)
Positive relationship quality, child [1,5]
 Wave 1 4.56 (0.58)
 Wave 2 4.61 (0.62)
 Wave 3 4.65 (0.53)
Negative relationship quality, child [1,5]
 Wave 1 2.44 (1.29)
 Wave 2 1.92 (1.16)
 Wave 3 1.90 (1.12)
Mental health (depressive symptoms) [0,57]
 Wave 1 10.28 (9.53)
 Wave 2 8.01 (8.86)
 Wave 3 8.21 (8.66)
Physical (self-rated) health [1,5]
 Wave 1 4.01 (0.93)
 Wave 2 3.79 (0.98)
 Wave 3 3.70 (0.96)
Open in a new tab

As seen in Table 4 and Figure 5, respondents who worked then retired experienced a decline in the geographic proximity of their network members (i.e., decrease in the percentage of network members living within a one-hour drive) at a greater rate than those who worked continuously. Those who retired experienced about a one percentage decrease every two years. Among those who worked continuously over the same time period, geographic proximity stayed relatively stable (see Figure 5). No link was found between work trajectory and change in network size.

Table 4.

Effect of work to retirement transition on network size and geographic proximity over time

Network size Network Geographic Proximity
b (SE) b (SE)
Intercept 7.16*** (0.54) 79.17*** (2.72)
Age (W1, centered at 40) −0.02 (0.01) −0.016
Years of education (W1, centered at 12) 0.21** (0.08) −1.99*** (0.38)
White 1.65*** (0.39) −1.76 (1.96)
Female 2.12*** (0.35) 0.74 (1.75)
Married or lives with partner 1.28*** (0.28) 0.36 (1.39)
Worked then retired 0.59 (0.55) 6.82* (2.74)
Other work trajectory 0.16 (0.54) −1.04 (2.68)
Survey year 0.02 (0.02) −0.08 (0.09)
Survey year * Worked then retired  −0.05 (0.03) −0.46** (0.14)
Survey year * Other work trajectory −0.01 (0.03) −0.12 (0.14)
SD (Intercept) 3.48 18.43
SD (Survey year) 0.11 0.86
N-Level 1 1,723 1,723
N-Level 2 593 593
LRT Chi-squared (M0, M1)a 3.97 (13, 15) 11.73** (13, 15)
Open in a new tab

Notes:

*

p<0.05;

**

p<0.01;

***

p<0.001;

a

Likelihood ratio test Chi-squared statistic calculated by comparing models to alternative models without survey year × work interactions.

Figure 5.

Figure 5.

Open in a new tab

Network geographic proximity over time by work-retirement trajectory

DISCUSSION

In this paper we used three examples to illustrate how the Convoy Model lens provides unique opportunity to see the multidimensional and dynamic character of linked lives across time and space. Findings illustrate linked lives through multiple instances of social relationships and influenced by various contexts. Further, the consequences of linked lives for mental health are consistent across the life course while influence on physical health is variable. Below we consider how key concepts from the Convoy Model situate the ways in which linked lives form and function at various levels and across multiple contexts to have far reaching effects on life outcomes.

Consequences of linked lives on later life health

The Convoy Model identifies social network size as a key dimension of social relations, and hence a specific instance of linked lives. Importantly, indicating network size as the number of close and important others represents the nature of linked lives as defined by the individual. Linked lives defined as network size and closeness reveal important consequences on later life health. The finding that larger network size at Wave 1 was associated with better mental health at all three waves, spanning twenty-three years across age groups, suggests an impressive, consistent, and long-term effect of earlier life network size. This supports the hypothesis that social relations have a long-term influence that is consistent over the life-span (i.e., across age contexts). Studies have also examined this association from the opposite direction (i.e., the effect of mental health on network structure), and found that better mental health is predictive of having more friends (Negriff, 2019; Schaefer, Kornienko, & Fox, 2011). Therefore, it is important to acknowledge that some part of the association we observed may due to this reciprocal effect.

A second interesting corollary is the more nuanced finding concerning closeness. Neither inner or outer circle placement were associated with mental health but greater number of middle circle members was associated with better mental health in all three waves. Composition of network circles as reported in other studies (e.g., Antonucci & Akiyama, 1987; Perkins, Ball, Kemp & Hollingsworth, 2013) indicates that there is relatively little variation concerning who is placed in the inner circle (i.e., mostly immediate family members). Middle circle membership, however, often includes connections beyond immediate family members (Ajrouch, 2008). The middle circle may distinguish those with a more enriched social network from those with a relatively limited or restricted one.

Social Relations: Structure and Physical Health

While there were no overall effects of network size on physical health, an interaction between network size, age, and time was evident. It is to be expected that self-reported physical health would be influenced by age and/or time, but the interactions with earlier life social networks reported above suggest a level of detail that is especially interesting. For the youngest and oldest age groups, having larger networks was linked to stability (i.e., no decrease) in physical health over time. In contrast, among 25–39 year-olds, physical health decreased significantly over time regardless of earlier life network size, although among the mid-life group those with smaller networks in Wave 1 reported no decrease in physical health over time. Since people in young adulthood generally experience the best health of their life during this time, and chronic illnesses begin to emerge in middle age, this overall decline in physical health being unrelated to network size is understandable. One possible explanation for why people in mid-life with smaller networks show no decline in physical health is that those with larger networks in this mid-life period are actually more stressed. It may be that larger networks are more demanding in terms of needs for assistance (Antonucci, Akiyama & Lansford, 1998). In sum, network size and closeness illustrate two key ways in which lives are linked to differentially predict health across time.

Linked lives in relation to social position

In addition to network size and closeness, relationship type and quality illustrate important aspects of linked lives. The Convoy Model identifies relationship quality as a key dimension of social relations, providing a heuristic for examining how individual lives are influenced by others in context. The social positions of race and education were examined to investigate relationship quality in two critical relationships over time: spouse/partner and parent-child.

Findings show that race is a key context for predicting relationship quality among spouses/partners over time. Among Black individuals who are married/in a committed relationship and remain so over 23 years, relationship quality with spouse/partner becomes more positive. It may be that having overcome challenges and years of shared experiences result in overall marital/partnership positivity. This may be especially meaningful among Black couples given that marriage is highly valued (Lincoln, Taylor, & Jackson, 2008), yet marital rates are overall low, which is often attributed to larger societal constraint (Tucker & Mitchell-Kernan, 1995). While race is a key context for understanding relationship quality among spouses/partners over time, education level is a key context for explaining relationship quality among parents and children.

The benefits of education are demonstrated in that parents with a higher education report lower and decreasing levels of negativity with their child over time. Education is known to be a powerful lifetime resource yielding access to more opportunities and resources, thus reducing the circumstances of limited and constrained resources which are often associated with relationship conflict. In sum, the contexts of race and education vary in their effects on relationship quality. Findings show that how individuals experience the link of key relationships over time illustrate important aspects of linked lives.

Linked lives in relation to situational change

In addition to a focus on social position as context, we also examined role change as a key situational context for elucidating how individual lives are linked to others. Our study had the unique advantage of comparing people who retired with those who continuously worked over 23 years. Network size and geographic proximity indicated two ways in which lives are linked. Our findings indicate that continuing to work was associated with no changes in network size or proximity, whereas those who retired showed decreases in the geographic proximity of their social network. Above and beyond age, the role transition from work to retirement matters. This supports the Convoy Model’s emphasis on situational characteristics as critical factors influencing how lives are linked. It is important to consider context in terms of social roles when trying to understand developmental change in characteristics of those with whom we are linked. These findings are consistent with previous studies which found stability in network size following retirement (Mor-Barak et al., 1992) and support the argument that retirement is a time of social continuity (Cozijnsen, Stevens, & van Tilburg, 2010). Further, the finding of declining network proximity in retirement may be consistent with Cozijnsen and colleagues’ (2010) finding that more recent cohorts report continued ties with co-workers into retirement. People may be retaining these ties after undergoing a post-retirement move, thus resulting in these continued ties becoming more distant in terms of geography proximity.

While we focused only on the role of being a worker, future research is needed to examine how other social roles and transitions into and out of them can influence how lives are linked. For example, Antonucci and colleagues (2019) point to a developing body of research showing how health transitions or what Parsons (1951) referred to as the sick role can influence social relations. This relates to a point made prior that social relations and health almost certainly have reciprocal effects. While it was beyond the scope of the present study, future research using lifespan panel data can do so and help disentangle at which point(s) in the life course each has an influence on the other.

In sum, across the three illustrative examples, findings illustrate the complexities and outcomes of linked lives. We demonstrate that advancing the importance of linked lives may be achieved by applying the concepts described in the Convoy Model of Social Relations. The Convoy Model presents key concepts to situate how linked lives are formed and function at various levels and across multiple contexts to have far reaching effects on life outcomes.

Acknowledgements:

The authors would like to thank the Life Course Development Program at the Institute for Social Research for comments made on earlier versions of this paper and Simon Brauer for data analytic assistance. This work was supported by grants from the U.S. National Institutes of Health (R01MH046549 to T.C.A.; R01MH066876 to T.C.A.; R01AG045423 to T.C.A.; and K01AG062754 to N.J.W.).

Footnotes

Declaration of interest: None

REFERENCES

  1. Ajrouch KJ (2008). Social isolation and loneliness among Arab American elders: Cultural, social, and personal factors. Research in Human Development, 5(1), 44–59. [Google Scholar]
  2. Antonucci TC (2001). Social relations: An examination of social networks, social support, and sense of control. In Birren JE & Schaie KW (Eds.), Handbook of the psychology of aging (pp. 427–453). San Diego, CA:, Academic Press. [Google Scholar]
  3. Antonucci TC (1986). Social support networks: A hierarchical mapping technique. Generations, X(4), 10–12. [Google Scholar]
  4. Antonucci TC, Ajrouch KJ, Webster NJ, & Zahodne LB (2019). Social relations across the life span: Scientific advances, emerging issues, and future challenges. Annual Review of Developmental Psychology, 1, 313–336. [Google Scholar]
  5. Antonucci TC, & Akiyama H (1994). Convoys of attachment and social relations in children, adolescents, and adults. In Nestmann F & Hurrerlmann K (Eds.), Social networks and social support in childhood and adolescence (pp. 37–52). New York, NY: Walter de Gruyter. [Google Scholar]
  6. Antonucci TC, & Akiyama H (1987). Social networks in adult life and a preliminary examination of the convoy model. Journal of gerontology, 42(5), 519–527. [DOI] [PubMed] [Google Scholar]
  7. Antonucci TC, Akiyama H, & Lansford JE (1998). Negative Effects of Close Social Relations. Family Relations, 47(4), 379. [Google Scholar]
  8. Antonucci TC, Fiori KL, Birditt K, & Jackey LMH (2010). Convoys of social relations: Integrating life-span and life-course perspectives. In Lamb ME, Freund AM, & Lerner RM (Eds.), The handbook of life-span development, Vol. 2. Social and emotional development (pp. 434–473). John Wiley & Sons, Inc. [Google Scholar]
  9. Belle DE (1983). The impact of poverty on social networks and supports. Marriage and Family Review, 54, 89–103. [Google Scholar]
  10. Berkman LF, Glass T, Brissette I, & Seeman TE (2000). From social integration to health: Durkheim in the new millennium. Social science & medicine, 51(6), 843–857. [DOI] [PubMed] [Google Scholar]
  11. Birditt KS, Hartnett CS, Fingerman KL, Zarit SH, & Antonucci TC (2015). Extending the intergenerational stake hypothesis: Evidence of an intra-individual stake and implications for well-being. Journal of Marriage and Family, 77(4), 877–888. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Broman CL (1993). Race differences in marital well-being. Journal of Marriage and the Family, 724–732. [Google Scholar]
  13. Bryant CM, Taylor RJ, Lincoln KD, Chatters LM, & Jackson JS (2008). Marital satisfaction among African Americans and Black Caribbeans: Findings from the national survey of American life. Family relations, 57(2), 239–253. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Bulanda JR, & Brown SL (2007). Race-ethnic differences in marital quality and divorce. Social Science Research, 36(3), 945–967. [Google Scholar]
  15. Burnett K, & Farkas G (2009). Poverty and family structure effects on children’s mathematics achievement: Estimates from random and fixed effects models. The Social Science Journal, 46(2), 297–318. [Google Scholar]
  16. Carr D (2018). The linked lives principle in life course studies: classic approaches and contemporary advances. In Alwin DF, Felmlee DH, & Kreager DA (Eds.), Social Networks and the life course: integrating the development of human lives and social relations networks (pp. 41–63). Cham, Switzerland: Springer. [Google Scholar]
  17. Carstensen LL, Isaacowitz DM, & Charles ST (1999). Taking time seriously. A theory of socioemotional selectivity. The American psychologist, 54(3), 165–181. [DOI] [PubMed] [Google Scholar]
  18. Chappell NL, & Havens B (1985). Who helps the elderly person: A discussion of informal and formal care. Social bonds in later life, 211–227. [Google Scholar]
  19. Chen E, & Berdan LE (2006). Socioeconomic status and patterns of parent–adolescen interactions. Journal of research on adolescence, 16(1), 19–27. [Google Scholar]
  20. Cozijnsen R, Stevens NL, & van Tilburg TG (2010). Maintaining work-related personal ties following retirement. Personal Relationships, 17(3), 345–356. [Google Scholar]
  21. Daw J, Verdery AM, & Margolis R (2016). Kin count(s): Educational and racial differences in extended kinship in the United States. Population and Development Review, 42(3), 491–517. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Fingerman KL, Huo M, & Birditt KS (2020). Mothers, fathers, daughters, and sons: Gender differences in adults’ intergenerational ties. Journal of Family Issues, 41(9), 1597–1625. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Fiori KL, Antonucci TC, & Akiyama H (2008). Profiles of social relations among older adults: A cross-cultural approach. Ageing and Society, 28(2), 203–231. [Google Scholar]
  24. Fletcher JM (2014). Late life transitions and social networks: The case of retirement. Economics Letters, 125(3), 459–462. [Google Scholar]
  25. George LK (1993). Sociological perspectives on life transitions. Annual review of sociology, 19(1), 353–373. [Google Scholar]
  26. Granovetter MS (1973). The strength of weak ties. American journal of sociology, 78(6), 1360–1380. [Google Scholar]
  27. Holt-Lunstad J, Smith TB, & Layton JB (2010). Social relationships and mortality risk: a meta-analytic review. PLoS Med, 7(7), e1000316. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Howard JH, Marshall J, Rechnitzer PA, Cunningham DA, & Donner A (1982). Adapting to retirement. Journal of the American Geriatrics Society, 30(8), 488–500. [DOI] [PubMed] [Google Scholar]
  29. Huxhold O, Fiori KL, Webster NJ, & Antonucci TC (2020). The Strength of weaker ties: An underexplored resource for maintaining emotional well-being in later life. The Journals of Gerontology: Series B, 75(7), 1433–1442. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Kahn RL, & Antonucci TC (1980). Convoys over the Life Course: Attachment, Roles, and Social Support. In Baltes PB & Brim OC (Eds.), Life-span, Development, and Behavior (pp. 254–283). New York: Academic Press. [Google Scholar]
  31. Lachman ME, & Agrigoroaei S (2010). Promoting functional health in midlife and old age: Long-term protective effects of control beliefs, social support, and physical exercise. PLoS one, 5(10), e13297. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Lareau A (2002). Invisible inequality: Social class and childrearing in black families and white families. American sociological review, 747–776. [Google Scholar]
  33. Lin N (2001). Social Capital: A Theory of Social Structure and Action. Cambridge: Cambridge University Press. [Google Scholar]
  34. Lincoln KD, Taylor RJ, & Jackson JS (2008). Romantic relationships among unmarried African Americans and Caribbean Blacks: Findings from the national survey of American life. Family relations, 57(2), 254–266. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Litwak E, & Longino CF Jr (1987). Migration patterns among the elderly: A developmental perspective. The Gerontologist, 27(3), 266–272. [DOI] [PubMed] [Google Scholar]
  36. McPherson M, Smith-Lovin L, & Cook JM (2001). Birds of a feather: Homophily in social networks. Annual review of sociology, 27(1), 415–444. [Google Scholar]
  37. Mead GH (1934). Mind, self, and society: From the standpoint of a social behaviorist (Vol. 1). Chicago: University of Chicago Press. [Google Scholar]
  38. Mor-Barak ME, Scharlach AE, Birba L, & Sokolov J (1992). Employment, social networks, and health in the retirement years. The International Journal of Aging and Human Development, 35(2), 145–159. [DOI] [PubMed] [Google Scholar]
  39. Negriff S (2019). Depressive symptoms predict characteristics of online social networks. Journal of Adolescent Health, 65(1), 101–106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Palmore E, Cleveland WP, Nowlin JB, Ramm D, & Siegler IC (1979). Stress and adaptation in later life. Journal of Gerontology, 34(6), 841–851. [DOI] [PubMed] [Google Scholar]
  41. Pan X, & Chee KH (2020). The power of weak ties in preserving cognitive function: a longitudinal study of older Chinese adults. Aging & mental health, 24(7), 1046–1053. [DOI] [PubMed] [Google Scholar]
  42. Parsons T (1951). The Social System. New York: Free Press. [Google Scholar]
  43. Perkins MM, Ball MM, Kemp CL, & Hollingsworth C (2013). Social relations and resident health in assisted living: An application of the convoy model. The Gerontologist, 53(3), 495–507. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Portes A (1998). Social capital: Its origins and applications in modern sociology. Annual review of sociology, 24(1), 1–24. [Google Scholar]
  45. Preacher KJ, Curran PJ, & Bauer DJ (2006). Computational tools for probing interaction effects in multiple linear regression, multilevel modeling, and latent curve analysis. Journal of Educational and Behavioral Statistics, 31(4), 437–448. [Google Scholar]
  46. Radloff LS (1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Measurement, 1, 385–401. [Google Scholar]
  47. Schaefer DR, Kornienko O, & Fox AM (2011). Misery does not love company: Network selection mechanisms and depression homophily. American Sociological Review, 76(5), 764–785. [Google Scholar]
  48. Settersten RA Jr (2015). Relationships in time and the life course: The significance of linked lives. Research in Human Development, 12(3–4), 217–223. [Google Scholar]
  49. Tucker MB, & Mitchell-Kernan C (Eds.). (1995). The decline in marriage among African Americans: Causes, consequences, and policy implications. Russell Sage Foundation. [Google Scholar]
  50. Van Tilburg T (2003). Consequences of men’s retirement for the continuation of work-related personal relationships. Ageing International, 28(4), 345–358. [Google Scholar]
  51. Webster NJ, Ajrouch KJ, Wan WH, & Antonucci TC (2017). Environmental Context and Social Relationships: A Relational Perspective on Health Disparities. In Dick AS & Muller U (Eds.), Advancing Developmental Science: Philosophy, Theory, and Method (pp. 210–221). Psychology Press: New York. [Google Scholar]
  52. Webster NJ, Antonucci TC, Ajrouch KJ, & Abdulrahim S (2015). Social networks and health among older adults in Lebanon: The mediating role of support and trust. Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 70(1), 155–166. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Wrzus C, Hänel M, Wagner J, & Neyer FJ (2013). Social network changes and life events across the life span: a meta-analysis. Psychological bulletin, 139(1), 53–80. [DOI] [PubMed] [Google Scholar]

RESOURCES