Intergenerational Emotional Cohesion and Psychological Well-Being of Older Adults in Rural China: A Moderated Mediation Model of Loneliness and Friendship Ties

Abstract

Objectives

Although researchers have linked intergenerational emotional cohesion (IEC) to psychological well-being (PWB) among older adults, the mechanisms and conditions under which IEC is related to PWB—particularly in rural areas—are less well understood. This study analyzed data from rural China to examine whether loneliness mediated the relationship between IEC and PWB, and whether friendship ties moderated the strength of the direct and indirect relationships between IEC and PWB.

Methods

Mediation and moderated mediation models were tested using a sample of rural adults aged 60 and older (N = 958) from the Longitudinal Study of Older Adults in Anhui Province, China. Measures included IEC, friendship ties, loneliness, and 2 PWB indicators—depressive symptoms and life satisfaction.

Results

The results revealed that IEC was negatively related to loneliness, which in turn was associated with depressive symptoms and life satisfaction. Furthermore, this indirect pathway linking IEC and depressive symptoms (but not life satisfaction) was positively conditioned on the size of friendship ties.

Discussion

This study advances our understanding of the mechanism through which IEC influences PWB in older adults. Alleviating loneliness could help boost PWB. Other implications for practice and future research are discussed.

Aging parents, Depression, Friendship ties, Intergenerational relations, Loneliness

Depression is a major public health concern worldwide, particularly for older adults among whom the lifetime prevalence of mental illness is 15% (World Health Organization [WHO], 2017). Research shows that more than three of 10 (30.8%) adults aged 65 and older in rural China experienced mild or severe depression during the previous week (Gao et al., 2009). WHO’s promotion of strengthening subjective well-being as an avenue to combat mental health problems suggests that subjective well-being is an important domain of psychological health (WHO, 2001). Thus, it is imperative to examine these two dimensions of psychological well-being (PWB)—depressive symptoms and life satisfaction—as distinct representations of a shared underlying construct. Theories in family gerontology and empirical research suggest that support from adult children plays a significant role in protecting older adults’ PWB (Cong & Silverstein, 2015; Kahn & Antonucci, 1980; Lai et al., 2019). However, we know relatively little about the mechanisms by which receipt of this support benefits PWB in familistic cultures and societies. One mechanism by which intergenerational emotional cohesion (IEC) likely confers PWB benefits to older adults is by improving emotional states such as reducing loneliness (Hawkley & Cacioppo, 2010; Perese & Wolf, 2005). While extant research has shown that loneliness predicts poor mental health in later life (Cacioppo et al., 2006; Hawkley & Cacioppo, 2010; Park et al., 2019), the linkage among IEC, loneliness, and PWB of older adults has not been clearly drawn in the empirical literature.

Furthermore, IEC has rarely been contextualized within the broader social network of nonkin such as friends. Gerontological theories suggest that friendship ties may contribute to older adults’ PWB in a way that is unique to other social partners (Messeri et al., 1993; Uehara, 1994). Friends are typically age-peers and, unlike kin relations, are more often voluntarily chosen based on shared values and lifestyles. This suggests that friends are vital social connections and sources of support for older adults that may serve as substitutes for weak family support (Blieszner et al., 2019; Fiori et al., 2020; Litwak, 1985); alternatively, friends may be equally important as family and both may mutually enhance mental well-being among older individuals (Cantor, 1979). However, research on the confluence of kin and nonkin relationships has received scant attention.

This investigation examined whether IEC improves older adults’ PWB, whether the impact of IEC is, in part, indirectly explained by reductions in loneliness, and how friendship ties modify the direct and indirect benefits of IEC to PWB. That we studied these issues in rural China—a strongly familistic region experiencing rapid social change—put into relief the question of whether intergenerational relations still serve as a salient socioemotional resource for older adults.

The Context of Rural China

The current study is set in rural China, where two thirds of older adults are concentrated (National Bureau of Statistics of China, 2020). Intergenerational relationships in China have a strong basis in the Confucian norm of filial piety, a cultural value that emphasizes the responsibility of adult children to ensure the well-being of their older parents (Sung, 1995). Because filial piety tends to be stronger in rural China than in urban China, its abrogation has potentially serious implications for the well-being of older adults in rural regions. Research demonstrates that intergenerational support to older adults enhances the mental well-being of older rural parents, particularly those left behind in their villages by labor-migrant children (Chen et al., 2014; Fu & Ji, 2020; Guo et al., 2009).

Despite the fact that intergenerational solidarity is traditionally strong in rural China, loneliness is quite prevalent in rural areas of the country (Guo & Chen, 2009; Wang et al., 2011). Research showed that over one third of older rural Chinese individuals report being unable to find an intimate person to talk with, compared to only about one quarter of their urban counterparts (Guo & Chen, 2009). Furthermore, rapid social and economic development has pushed the younger generation to migrate to urban areas for better employment opportunities, which might have some dire effects on aging parents who are geographically distant from their adult children (Guo et al., 2009). In recent years, responsibility for older adults has begun to shift away from family in rural and urban regions of China (Cheung et al., 2006; Du, 2013). Thus, changes in the social and cultural context of rural China make it increasingly important to understand the mechanism and conditions under which IEC contributes to the PWB of older adults exposed to such changes.

Intergenerational Emotional Cohesion and Psychological Well-Being

Intergenerational relationships are important sources of informal social and emotional support for older people to maintain their emotional well-being (Chen & Silverstein, 2000; Guo et al., 2009; Merz & Huxhold, 2010; Silverstein & Bengtson, 1994). In this study, we rely on the intergenerational solidarity framework, the dominant paradigm in the aging literature to conceptualize intergenerational integration (Bengtson, 1975; Silverstein et al., 1997). IEC refers to emotional cohesion with adult children (also called affectual solidarity). It comprises feelings of affection, emotional closeness, and emotional support and forms the core dimension of the framework. Studies in Western countries have shown emotional cohesion to be positively consequential for the mental health of older parents (Kahn & Antonucci, 1980; Lai et al., 2019). Similar results to the above have been found in research on older families in China with respect to a wide range of mental health and well-being outcomes such as depressive symptoms (Chen et al., 2014; Fu & Ji, 2020), life satisfaction (Chen & Jordan, 2016; Peng et al., 2019), and PWB (Chen & Silverstein, 2000; Silverstein et al., 2006). However, little research to date has examined psychological pathways that mediate the association between IEC and older parents’ PWB in China and other familistic societies.

Loneliness as a Mediator

Loneliness is an emotionally painful and distressing condition that has been documented as a public health epidemic in modern times (Andersson, 1998; Cacioppo et al., 2006). Loneliness encompasses deficiencies in both subjective quality and objective quantity of interpersonal relationships (Andersson, 1998; Heinrich & Gullone, 2006). Prevalence of loneliness is particularly high among older adults (Cohen, 2000) and represents a risk factor for mental well-being because it enhances feelings of vulnerability in an already marginalized population (Hawkley & Cacioppo, 2010). The lack of meaningful social connections that induce loneliness is a well-documented social condition in later life that predicts mental health problems such as depressive symptoms (Hawkley & Cacioppo, 2010; Perese & Wolf, 2005). Empirically, loneliness has consistently been found to increase the risk of depressive symptoms in Western (Cacioppo et al., 2006) and Chinese samples (Wu et al., 2011).

Rewarding familial relationships can make people less prone to loneliness (Hawkley & Cacioppo, 2010), yet there is limited non-Western research that has focused on the emotional aspects of parent–adult child relationships, which are found to reduce loneliness and subsequent mental health problems (De Jong Gierveld et al., 2012). Although limited, studies found that loneliness partially mediated the relationship between family support networks and depressive symptoms among older Koreans living in another highly familistic culture (Park et al., 2013, 2019). The loneliness-mediated effects of IEC on PWB follow the logic of Berkman et al.’s (2000) conceptual model positing that social relationships influence PWB through the psychological mechanism of which loneliness is one component that links supportive relationships to PWB. Taken together, we investigated the mediational effects of loneliness in linking the relationships between IEC and older adults’ PWB in a rural Chinese context, where filial support is apt to be particularly salient.

Friendship Ties as a Moderator of Intergenerational Effects

Friendship ties function as another source of support that significantly contributes to older adults’ mental well-being (Fiori et al., 2020). Relational connectedness and support provided by friends offer meaning to older adults’ lives, which can, in turn, improve mood and provide a sense of well-being (ten Bruggencate et al., 2018). However, it is worth noting that most studies regarding friendship have been conducted in Western societies, leaving it unclear whether friendships are as important in different sociocultural contexts (Dilworth-Anderson & Marshall, 1996). The few existing studies examining the relationship between friendship ties and PWB of Chinese older adults have reported mixed findings, further calling into question the generalizability of these conclusions. For example, some studies in China have demonstrated that friendship ties were negatively related to depressive symptoms (Tang et al., 2020) and positively associated with PWB (Lei et al., 2015; Li & Zhang, 2015), whereas other studies have shown no such relationships (Chen et al., 2014).

The simultaneous impact of kin and nonkin support on the well-being of older adults has received little attention in China. International research relying on the convoy model of social relationships (Antonucci & Akiyama, 1995) suggests that both children and friends play important roles in protecting older adults’ PWB. However, hierarchical theories of primary groups suggest differential effects depending on the role that friends and adult children play with respect to each other in the social networks of older adults. Specifically, the hierarchical-compensatory theory posits that nonkin relationships are important contributors to older adults’ PWB, only when kin relations are not available (Cantor, 1979)—what we label as substitution. Alternatively, the task-specific theory proposes that both types of relationships are complementary to each other because each fulfills a different function and the presence of both is necessary for optimal PWB (Litwak, 1985)—what we label as reinforcement. Given the integration of diverse social relationships in the communal tight-knit rural villages of China, these two sources of support are likely to be mutually reinforcing because family members and friends of older adults are likely to know and interact with each other (Huxhold et al., 2014). Thus, given the social ecology of rural China, we hypothesize that reinforcement, rather than substitution, will characterize the interdependence between family and friendship relationships in relation to PWB.

Research Questions and Hypotheses

This study examines a key psychological mechanism by which IEC relates to PWB and how wider social relationships condition this mechanism in a sample of older adults from rural China. Specifically, we examine the following two research questions (see Figure 1 for the conceptual model). First, does loneliness mediate the relation between IEC and PWB (i.e., depressive symptoms and life satisfaction)? It is hypothesized that IEC will be negatively associated with loneliness (Path A1), and loneliness will be positively associated with depressive symptoms (Path A2) and negatively related to life satisfaction (Path A3).

Figure 1.

Conceptual model linking intergenerational emotional cohesion to depressive symptoms and life satisfaction. Note: A Paths are links for the mediation model; B Paths represent interaction effects—for example, B1 represents the interaction effect between intergenerational emotion cohesion and friendship ties on older adults’ depressive symptoms. The model also controlled age, gender, marital status, education, self-reported health, functional limitations, number of chronic diseases, living arrangements, number of children, and log financial support.

Second, we examine whether friendship ties moderate the direct and loneliness-mediated indirect links between IEC and depressive symptoms and life satisfaction—namely, the link between IEC and depressive symptoms and life satisfaction (Path B1 and B2) and the link between loneliness and depressive symptoms and life satisfaction (Path B3 and B4). Given that loneliness is a negative subjective experience and does not necessarily have to do with the amount of social interactions (Lykes & Kemmelmeier, 2014), we do not include friendship ties as a moderator of the relation between IEC and loneliness. Considering the two competing theories of social networks in later life (i.e., substitution vs. reinforcement) in the context of the aforementioned cultural and social context of rural China, we propose that the reinforcement hypothesis will characterize the moderation model proposed; the association of IEC with PWB will be stronger when older adults have stronger friendship ties.

Method

Data and Participants

Data were derived from the Longitudinal Study of Older Adults in Anhui Province. Anhui Province, a less developed province located in China’s eastern region, was selected due to its high labor migration rates and large shares of the rural population. The baseline sample was recruited in 2001 using a stratified, multistage, random-sampling methodology (Guo et al., 2009). Follow-up data were collected in 2003, 2006, 2009, 2012, 2015, and 2018, and three replenishment subsamples were added in 2009, 2015, and 2018 to rebalance the age distribution of the sample. This study focused on the 2018 wave data because loneliness, one of our main constructs of interest, was first measured in this wave. The full 2018 sample consisted of 1,234 respondents, including 319 survivors from the original sample and 915 from the replenishment samples. Nonmortality attrition was less than 10% across consecutive waves (Cong & Silverstein, 2015). Compared to nonsurvivors, survivors were younger (p < .001), more educated (p < .01), and more likely to be married (p < .001). There were no significant differences in gender or overall health between survivors and nonsurvivors.

This study’s analytic sample consisted of 958 older adults, ranging in age from 60 to 96 years (M_age = 70.10, SD = 7.31; 52% male), who had at least one living child and had valid data at the 2018 wave. We first removed 92 cases with missing data or no children. To avoid response bias due to cognitive impairment, we further removed 184 respondents who did not pass the standardized cognitive screening questions (Tang et al., 2020; Wang et al., 2014).

Measures

Dependent variables

PWB was operationalized by measures of depressive symptoms and life satisfaction, two distinct, but highly correlated variables, with depressive symptoms reflecting short-term emotional distress and life satisfaction reflecting a subjective cognitive evaluation of current quality of life (Diener, 1984). Depressive symptoms were measured by a nine-item adapted version of the Center for Epidemiologic Studies—Depressive symptoms scale (Radloff, 1977), which has established psychometric properties for Chinese older adults (Cheng & Chan, 2005). Respondents rated each item with one of the three responses (0 = rarely or none of the time, 1 = some of the time, 2 = most of the time). Responses were summed to create a depressive symptoms scale ranging from 0 to 18, with higher scores indicating more depressive symptoms. The reliability coefficient alpha for this scale was 0.79. Life satisfaction was assessed using seven items adapted from the Satisfaction With Life Scale (Diener et al., 1985), which asked respondents whether they agreed or disagreed with statements indicating contentment and discontentment with their current lives (e.g., “I am satisfied with life”). Responses were coded as 1 (agree) and 0 (disagree) and summed to create a life satisfaction scale, ranging from 0 (least satisfied) to 7 (most satisfied). The reliability coefficient alpha for this scale was 0.82.

Predictors

IEC was assessed using three items adapted from the affectual solidarity subscale of the Intergenerational Solidarity Inventory (Mangen et al., 1988). An example item is “Taking everything into consideration, how close do you feel to (this child)?” Questions were rated on a 3-point scale ranging from 0 (not at all) to 2 (very). An additive score was calculated for each child, ranging from 0 to 6. The reliability coefficient alpha for this scale ranged from 0.80 to 0.86 based on the children’s birth order. To create a summary score of IEC, we averaged scale scores across all children.

Loneliness was measured with six items from the eight-item UCLA Loneliness Scale short form (ULS-8; Hays & DiMatteo, 1987). Six items were selected based on the analysis of the reliability and validity of the Chinese version of the ULS-8 in rural older adults (Zhou et al., 2012). Respondents were asked how often they felt that they: “lack companionship,” “there is no one I can turn to,” “people are around me but not with me,” “left out,” “isolated from others,” and “unhappy being so withdrawn.” Respondents rated each item with one of the following responses (0 = never, 1 = rarely, 2 = sometimes, 3 = always). Responses were summed to create a loneliness scale ranging from 0 (least lonely) to 18 (most lonely). The reliability coefficient alpha for this scale was 0.87.

Friendship ties were assessed using the Friends subscale from Lubben’s Social Network Scale (LSNS; Lubben et al., 2006). An example item is “How many of your friends do you see or hear from at least once a month?” For each question, responses ranged from 0 to 5 (0 = 0, 1 = 1, 2 = 2, 3 = 3 or 4, 4 = 5–8, and 5 = 9 and above). As recommended by Lubben et al., responses were summed to create a friendship ties scale, ranging from 0 (weakest ties) to 15 (strongest ties). Although the friendship ties measure combines the size and intimacy of friendship networks, we use the weak-to-strong terminology to summarize both aspects of friendships in the aggregate. The reliability coefficient alpha for this scale was 0.87.

Covariates

A wide array of factors known to contribute to PWB were controlled when testing our hypotheses. We considered sociodemographic characteristics and health-related variables as covariates, including gender, education, marital status, financial support received from children, living arrangement (see Table 1 for more detailed coding), number of chronic diseases, self-rated health, and functional limitations (see Covariates in Supplementary Material for more details).

Table 1.

Descriptive Statistics and Correlations of the Studied Variables (N = 958)

Variable	1	2	3	4	5	6	7
1. IEC	1.00
2. Friendship ties	0.18**	1.00
3. Loneliness	−0.27**	−0.17**	1.00
4. Depressive symptoms (2018 wave)	−0.26**	−0.17**	0.57**	1.00
5. Life satisfaction (2018 wave)	0.25**	0.16**	−0.55**	−0.64**	1.00
6. Depressive symptoms (2015 wave)	−0.23**	−0.08*	0.26**	0.32**	−0.29**	1.00
7. Life satisfaction (2015 wave)	0.27**	0.12**	−0.24**	−0.26**	0.34**	−0.64**	1.00
8. Age	−0.09**	−0.07*	0.14**	0.17**	-0.06	0.08*	0.02
9. Female	−0.04	0.01	−0.07*	−0.17**	0.10**	−0.12**	0.12**
10. Married	0.03	0.07*	−0.26**	−0.19**	0.12**	−0.09*	0.05
11. Some schoolinga	0.03	0.06	−0.05	−0.18**	0.13**	−0.15**	0.16**
12. Self-reported health	0.17**	0.10**	−0.37**	−0.47**	0.40**	0.24**	0.18**
13. Functional limitations	−0.10**	−0.04	0.29**	0.45**	−0.35**	0.22**	−0.20**
14. Number of diseases	−0.08*	−0.03	0.16**	0.20**	−0.17**	0.16**	−0.10**
Living arrangementsb
15. Live alone	−0.01	−0.07*	0.20**	0.12**	−0.09**	0.05	−0.01
16. Live with spouse only	0.02	−0.01	−0.11**	−0.07*	0.04	−0.07*	0.06
17. Others	−0.03	0.03	0.04	0.07*	0.03	0.07*	−0.01
18. Live with others only	0.01	0.02	0.04	0.08*	−0.02	0.01	−0.05
19. Number of children	0.01	0.02	0.02	0.08*	−0.01	0.001	0.09*
20. Log financial support	0.24**	0.16**	−0.10**	−0.13**	0.10**	−0.08	0.12**
M	5.22	4.71	3.69	4.84	5.17	5.29	5.01
SD	1.11	3.80	3.85	3.65	2.09	3.50	1.99

Note: IEC = intergenerational emotional cohesion; SD = standard deviation.

^aReference group: no schooling.

^bWe created four dummy variables for living arrangement and percentages were shown in parentheses: live alone (11.6%); live with spouse only (41.3%); others = other living arrangements including those who live alone or live with others and have at least one child living in the same village (15.3%); live with others only = live with others only and have no children living in the same village (4.5%); reference group: live with children (27.3%).

**p < .01, *p < .05.

Analysis Plan

All continuous variables were standardized before testing the conceptual model (Figure 1). The conceptual model was tested using structural equation modeling with Mplus Version 8 (Muthėn & Muthėn, 2017). We used the full information maximum likelihood estimation method, which enables the full usage of all available data (Muthėn & Muthėn, 2017). Bootstrapping with maximum likelihood discrepancy was used to establish the statistical significance of indirect effects and bias-corrected 95% confidence intervals (CIs; Mackinnon et al., 2004; Preacher & Hayes, 2008). As a robustness check to account for possible endogeneity of the main predictor variables with regard to PWB, we included depressive symptoms and life satisfaction from an earlier wave (2015) as lagged controls.

First, the main mediation paths in the model (only A paths) were tested. The indirect effects from emotional cohesion to depressive symptoms and life satisfaction were tested using the delta method described by Sobel (1982). Second, the moderated mediation model (both A paths and B paths) was tested. When an interaction effect was significant, simple slope analyses (1 SD above and below the mean of the moderator) were conducted (Aiken & West, 1991). A set of control variables described above were included as covariates.

Results

Descriptive and Correlational Analyses

The large majority of respondents were married (80%), and approximately half had no formal education (55%). In terms of living arrangements, 27% of the sample lived with their children. Almost two thirds of older adults reported having no difficulties performing any tasks corresponding to activities of daily living (64%; M = 2.12, range = 0–29). The median and mode of self-reported health are 2 (fair). Table 1 displays descriptive statistics and bivariate correlations of the study variables. As expected, we found that the relationships among all variables in the conceptual model were all significant and in the expected direction. Overall, these findings indicated that older adults with stronger IEC had lower levels of loneliness, fewer depressive symptoms, greater life satisfaction, and stronger friendship ties. Notably, older adults in this sample reported, on average, weak friendship ties (M = 4.71). According to Lubben et al. (2006), individuals with scores less than 6 on the three-item LSNS-6 Friends subscale are considered to have marginal friendship ties.

Path Analysis

Mediation analyses

First, the mediation model with all the A paths in Figure 1 was tested. The model fit the data well, χ ² (16) = 77.27, p < .001, comparative fit index (CFI) = 0.96, root mean square error of approximation (RMSEA) = 0.06 [0.05–0.08], standardized root mean square residual (SRMR) = 0.03. We found that stronger IEC reduced loneliness (β = −0.20, p < .001); and greater loneliness increased depressive symptoms (β = 0.40, p < .001) and reduced life satisfaction (β = −0.43, p < .001). The 95% bootstrapped CIs sample revealed that the indirect effects from IEC to PWB outcomes through loneliness was significant (depressive symptoms: β = −0.08, 95% CI [−0.11 to −0.05]; life satisfaction: β = 0.09, 95% CI [0.05–0.12]). Thus, loneliness significantly mediated the relationship between IEC and PWB.

Moderated mediation analyses

The moderated mediation model shown in Figure 1 was tested with the addition of interaction terms. The overall model fit the data well, χ ² (4) = 9.04, p = .06, CFI = 0.99, RMSEA = 0.04 (0.00–0.07), SRMR = 0.01. Results revealed that the indirect influence of IEC on depressive symptoms through loneliness was moderated by the strength of friendship ties (Table 2). Specifically, the indirect effect was stronger for older adults with stronger friendship ties (β = −0.09, 95% CI [−0.13 to −0.06]) than for older adults with weaker friendship ties (β = −0.07, 95% CI [−0.10 to −0.04]). The significant mediation and interaction effects suggested that both the direct effect of IEC on older adults’ depressive symptoms and the indirect effect from IEC to older adults’ depressive symptoms through loneliness were significantly larger for those who had stronger friendship ties. This result supported the reinforcement hypothesis proposing that the benefits to PWB of having closer kinship ties are enhanced by stronger friendship ties—in this instance by reducing loneliness.

Table 2.

Coefficients for Moderated Mediation Model

	Loneliness		Depressive symptoms		Life satisfaction
	Direct effects		Direct effects		Direct effects
	β (SE)	95% CIa	β (SE)	95% CIa	β (SE)	95% CIa
Predictors
IEC	−0.19 (0.03)	−0.27, −0.12	−0.08 (0.03)	−0.13, −0.02	0.10 (0.03)	0.03, 0.16
C1 Age	−0.01 (0.04)	−0.08, 0.07	−0.03 (0.03)	−0.09, 0.04	0.08 (0.03)	0.001, 0.15
C2 Female	−0.01 (0.03)	−0.08, 0.05	−0.04 (0.03)	−0.10, 0.01	−0.01 (0.03)	−0.05, 0.06
C3 Married	−0.19 (0.05)	−0.29, −0.09	0.01 (0.04)	−0.07, 0.08	−0.01 (0.04)	−0.09, 0.06
C4 Some schoolingb	−0.01 (0.03)	−0.05, 0.08	−0.10 (0.03)	−0.15, −0.05	0.08 (0.03)	0.02, 0.13
C5 Self-reported health	−0.25 (0.03)	−0.31, −0.18	−0.19 (0.03)	−0.24, −0.13	0.15 (0.03)	−0.09, 0.06
C6 Functional limitations	0.13 (0.05)	0.04, 0.22	0.22 (0.03)	0.16, 0.28	−0.17 (0.03)	−0.24, −0.10
C7 Number of chronic diseases	0.01 (0.03)	−0.05, 0.07	0.00 (0.03)	−0.04, 0.05	−0.01 (0.02)	−0.06, 0.04
Living arrangementsc
C8 Live alone	0.07 (0.04)	−0.02, 0.16	0.03 (0.03)	−0.03, 0.10	0.01 (0.03)	−0.05, 0.08
C9 Live with spouse only	−0.01 (0.03)	−0.07, 0.05	0.03 (0.03)	−0.04, 0.07	0.02 (0.03)	−0.04, 0.09
C10 Othersd	−0.02 (0.04)	−0.08, 0.05	0.02 (0.03)	−0.04, 0.08	0.07 (0.03)	0.01, 0.13
C11 Live with others onlye	−0.05 (0.04)	−0.12, 0.03	0.05 (0.03)	−0.01, 0.11	0.01 (0.03)	−0.07, 0.08
C12 Number of children	−0.03 (0.04)	−0.10, 0.03	0.03 (0.03)	−0.03, 0.08	−0.01 (0.03)	−0.09, 0.05
C13 Log financial support	−0.03 (0.03)	−0.10, 0.04	−0.03 (0.03)	−0.08, 0.03	−0.00 (0.04)	−0.06, 0.07
FT	—	—	−0.02 (0.03)	−0.10, −0.00	0.04 (0.03)	−0.01, 0.10
Mediator
Loneliness	—	—	0.41 (0.03)	0.35, 0.46	−0.43 (0.03)	−0.49, −0.37
Interaction effects
IEC × FT	—	—	−0.07 (0.03)	−0.12, −0.04	0.02 (0.03)	−0.02, 0.08
Loneliness × FT	—	—	0.06 (0.03)	0.01, 0.11	−0.03 (0.03)	−0.09, 0.02
R2	0.24	—	0.48	—	0.39	—
Indirect effects	—	—	−0.08	−0.11, −0.05	0.08	0.05, 0.12

Note: IEC = intergenerational emotion cohesion; FT = friendship ties.

^aBootstrap estimate and 95% bias-corrected confidence interval (CI) was based on 5,000 bootstrap samples. CI does not contain 0, indicating there is a statistically significant result. Significant results are denoted in bold.

^bReference group: no schooling.

^cReference group: live with children.

^dOthers = other living arrangements including those who live alone or live with others and have at least one child living in the same village.

^eLive with others only = live with others only and have no children living in the same village.

In order to depict the results in a more descriptive manner, simple slope analysis for IEC was conducted at strong versus weak friendship ties (1 SD above vs. below the sample mean) to decompose the interaction effect (Aiken & West, 1991). As shown in Figure 2, IEC was negatively related to depressive symptoms (β = −0.16, p < .001) when older adults reported strong friendship ties. In contrast, when older adults reported having weak friendship ties, IEC was unrelated to older adults’ depressive symptoms (β = −0.02, p = .48). The results of simple slope analyses of loneliness on depressive symptoms indicated that the effect between loneliness and depressive symptoms was stronger for older adults with stronger friendship ties (β = 0.47, p < .001; Figure 3) than those with weaker friendship ties (β = 0.35, p < .001). For life satisfaction, there was no significant moderating effect of friendship ties on the direct link between IEC and life satisfaction or on the second segment of the mediating relationship between loneliness and life satisfaction. These results did not support the proposed moderated mediation model for life satisfaction.

Figure 2.

Interaction effect between older adults’ intergenerational emotional cohesion and friendship ties on their depressive symptoms. Note: Older adults’ intergenerational emotional cohesion and depressive symptoms were negatively associated with each other when older adults’ friendship ties were strong (1 SD above the sample mean) but not significantly correlated when friendship ties were weak (1 SD below the sample mean).

Figure 3.

Interaction effect between older adults’ loneliness and friendship ties on their depressive symptoms. Note: The link between loneliness and depressive symptoms was stronger when friendship ties were stronger (1 SD above the sample mean) than when friendship ties were weaker (1 SD below the sample mean).

To test the robustness of our results with respect to the direction of influence, we also estimated the model shown in Figure 1 controlling for lagged versions of each dependent variable as measured in the 2015 survey. Lagged control variables had acceptable measurement properties (reliability coefficient alpha was 0.76 for depressive symptoms and 0.77 for life satisfaction). Analyses with lagged controls (not shown) yielded similar results to those in analyses without lagged controls (Supplementary Figure 1 and Table 1).

Discussion

Despite scholars having long proposed the central role that IEC plays in older adults’ mental well-being, the mechanism and conditions under which IEC relates to PWB remain understudied, especially in familistic cultures. To address this gap, we first examined whether loneliness mediated the relation between IEC and older adults’ PWB. We then tested the hypothesis that friendship ties moderate both the direct and indirect (through loneliness) effects of IEC on older adults’ PWB (i.e., depressive symptoms and life satisfaction). Consistent with our hypotheses, results demonstrated that stronger IEC was related to reduced depressive symptoms and increased life satisfaction directly and indirectly through reducing feelings of loneliness. The indirect pathway linking IEC to fewer depressive symptoms through loneliness was more evident when older adults reported stronger friendship ties, suggesting reinforcement rather than substitution between these two types of social ties. Friendship ties did not moderate direct or indirect links between IEC and life satisfaction.

Consistent with past studies, we found significant associations between IEC, loneliness, depressive symptoms, and life satisfaction (Cacioppo et al., 2006; Merz & Huxhold, 2010; Park et al., 2019; Silverstein et al., 2006), supporting the intergenerational solidarity model of family functioning (Bengtson, 1975). High levels of emotional cohesion with adult children may provide aging parents a sense of family unity and continuity despite large geographic distances between generations, such as left-behind older parents. The primacy of family life appears to be robust in this traditional region of China, where filial piety is still strong.

This study sheds light on the underlying mechanism linking IEC and depressive symptoms and life satisfaction through loneliness. Consistent with attachment theory (Bowlby, 1982), IEC is associated with reduced loneliness. In addition, loneliness is positively related to depressive symptoms and negatively correlated with life satisfaction, in line with Cacioppo et al. (2006), who propose that loneliness is accompanied by the surveillance of social threat which induces negative mood and low life satisfaction. Incorporating these theoretical notions, this study provides an inclusive understanding of how IEC relates to older adults’ PWB in the traditional context of rural China. Our findings that the effects of IEC on PWB are both direct but indirect through loneliness have important clinical and interventional implications. Considering that IEC and loneliness have been identified as predictors of older adults’ PWB (Cacioppo et al., 2006; Silverstein et al., 2006), special attention should be paid to alleviating loneliness in this population. Given that feeling lonely predicts depressive symptoms, a simple assessment of one’s emotional perception of being socially isolated could provide a marker for the risk of depression.

Another contribution of this study is in its consideration of friendship ties, a potentially critical but understudied source of support contributing to older adults’ PWB (Fiori et al., 2020). As expected, we find that friendship ties significantly moderate the direct path from IEC to depressive symptoms. Specifically, IEC is negatively related to depressive symptoms when friendship ties are relatively strong (but not relatively weak), suggesting that older adults with strong friendship ties benefit more from IEC. This finding suggests that IEC and friendship ties are mutually reinforcing with respect to reducing depressive symptoms of older adults in a rural region of China. That the protective effect of IEC on depression is enhanced by friendship ties might be due to the rapid shift away from relying solely on intergenerational family support among older adults in rural China and the emergence of more complex relational dynamics.

The modifying role of friendship ties in the strength with which loneliness reduces depressive symptoms suggests that friendship ties can protect older adults from the potential harmful effects of social vulnerability on depressive symptoms. This finding supports Cohen and Wills’ (1985) stress-buffering model in that friendship ties can act as an important coping resource that buffers the impact of feeling lonely on depression. In general, this finding has practical implications for the design of interventions that allow friendships to flourish in rural areas. Furthermore, our findings have cultural implications as support for older adults has begun to shift away from traditional family sources to new relational forms. Surprisingly, we did not find that friendship ties moderated the direct link between IEC and life satisfaction or their indirect link through loneliness. This suggests that relationships with adult children are consistently related to the quality of daily life regardless of the wider social environment.

Limitations

Several limitations of this study warrant attention. First, it should be acknowledged that all the measures in this study are self-reported and thus may be subject to biased reporting. However, we applied a standard clinical and epidemiological approach to selecting included data to avoid response bias due to cognitive impairment. Second, IEC, loneliness as well as depression, and life satisfaction were measured in the same wave (2018) due to data limitations. However, when we controlled for depressive symptoms and life satisfaction from an earlier wave, no substantial differences in estimates were found, thus adding to confidence in our results. Nevertheless, it should also be noted that this study is not formally prospective in terms of theoretically relevant predictors, limiting our ability to make causal inferences and suggesting that caution be taken in attributing causality to our results.

Third, owing to the data restrictions, we considered only positive aspects of intergenerational relationships and friend relationships. Prior studies have found that relational strain, conflict, and ambivalence exacerbate poor mental health, including depressive symptoms, anxiety, and mood disorders (Bertera, 2005; Mavandadi et al., 2007). Future studies should consider both positive and negative dimensions of relationships as both protective and risk factors in contributing to older adults’ PWB. Fourth, this study examined a unique sample of older adults from the same rural area, thereby limiting generalizability of our findings. Findings might vary across different sectors of China, most notably between rural and urban regions. For example, the observed link from IEC to PWB might be weaker among older adults in urban areas of China because the traditional cultural norm of filial piety is more dominant in rural areas. Research has documented that support from children plays a more influential role in PWB among rural adults, while support from neighborhood friends plays a more influential role among urban adults (Tang et al., 2020; Zhang et al., 2021).

Conclusions

This study uniquely considered both intergenerational and friendship relationships in relation to loneliness and PWB within the familistic culture of rural China. In doing so, we hope to have better captured the role of relational complexity in understanding the PWB consequences of these close relationships in later life. Emotional cohesion with adult children still matters for older parents’ PWB in this strongly filial culture, but its implication for parental depression appears to be contingent on friendship ties. This synergism between family and nonfamily ties may be a property of tight-knit rural villages that remain highly integrated across relational boundaries.

Our results point to the importance of loneliness as a modifiable risk factor for depression and poor life satisfaction that may be amenable to intervention. We hope our findings help stimulate the development of programs that reduce loneliness by strengthening social relations as a means of improving the mental well-being of older adults in rural China. Such programs can take the form of public senior centers that provide structured activities for older adults in service-deprived rural areas. Addressing the social antecedents of well-being among older adults in less developed rural regions of the world may produce the widest possible gains for quality of life in these vulnerable populations experiencing rapid social and family change.

Acknowledgments

We would like to thank Shuzhou Li and his research team for developing the data collection infrastructure for this project. Data, analytic methods, and study materials will be provided on reasonable request to the first author. This study was not preregistered.

Funding

This work was supported by grants from the National Natural Science Foundation of China (71573207, 72074177) and the Fogarty International Center of the National Institutes of Health (R03 TW01060).

Conflict of Interest

We have no conflict of interest regarding the publication of this article.

Author Contributions

X. Zhang designed the study, prepared the data set, performed all statistical analyses, interpreted the findings, and drafted the manuscript. M. Silverstein assisted with the study design, supervised the data analysis, and interpreted the results. Both contributed to revising the manuscript for submission.