Who We Live With and How We Are Feeling: A Study of Household Living Arrangements and Subjective Well-Being Among Older Adults in China

Abstract

Using a sample of Chinese adults over the age of 50 from wave 1 of the WHO Study on Global Ageing and Adult Health (n = 13,367), we investigated the relationship between living arrangements and subjective well-being (SWB) in regard to life satisfaction, happiness, and control. We also looked at the moderating role of resources, proxied by income and hukou status. Multivariate regression results indicate that living only with a spouse was significantly associated with better SWB. Multigenerational living arrangements may not always promote SWB, particularly when resources are constrained. Yet, results also underscore the importance of daughters and daughters-in-law in promoting SWB among older adults. Older adults in rural areas had better SWB, including greater life satisfaction if living with grandchildren only, compared to their urban peers living with a spouse only. Findings suggest that context matters in the association between living arrangements and older adults’ SWB.

Keywords

living arrangements older adults China subjective well-being WHO SAGE study

The “greying” of populations around the globe necessitates our understanding of the well-being of people aged 50 and above. This issue is important not only in its own right but also in its implications for population health. The increasing number of people aged 50 or older and the rise in the average age may bring new challenges to mid- and late-life as people live longer and have fewer children, on average, to provide support (Feng et al., 2012). While objective measures such as income and assets can help explain individual well-being, scholarship over the past 3 decades has increasingly acknowledged the value of subjective well-being (SWB). In other words, how people feel about how they are doing is as important as, if not more important than, objective measures (Ferring & Boll, 2010). Subjective well-being also is an important indicator of social system functioning, carrying important implications for public health. Prior scholarship has established a close link between psychological well-being and physical well-being: Feeling good tends to promote better physical health (Lin et al., 2003; Ohrnberger et al., 2017).

We contribute to the SWB scholarship on a crucial socio-cultural factor, living arrangements. Recent studies using Western samples have shown that living alone is significantly associated with loneliness, with implications for older adults’ health, SWB, and cognition (Hawkley et al., 2019; Suanet & van Tilburg, 2019). We add to this line of research using a nationally representative sample from China collected by the World Health Organization, the Study on Global Ageing and Adult Health (SAGE). We use a detailed categorization of living arrangements to conduct a nuanced investigation of the links between types of households and older adults’ SWB.

We focus on China due to both its large aging population and the significant economic and social changes taking place there in recent decades. As of 2016, China had 231 million adults aged 60 and above, 17% of China’s total population and the highest absolute number of older adults in the world (Ministry of Civil Affairs, 2017; United Nations, 2019). By 2053, the 60+ population is projected to grow to 487 million, accounting for 35% of China’s total population (Ge, 2014). Furthermore, since the 1980s, China’s economy has transformed rapidly, away from a primarily agrarian society with a planned economy to an urban, market-oriented society. Simultaneously, young generations from both urban and rural areas have migrated in historic numbers to more developed regions and cities in pursuit of economic opportunities and better lives (Wu et al., 2010). Their aging parents are often left behind, sometimes caring for grandchildren.

Socially and demographically, the one-child policy depressed China’s fertility rate from 4.77–6.11 before 1975, to 1.18 in 2010 (Guo, 2016; United Nations, 2019). The first generation to begin childbearing during the one-child policy era is now moving into late life (age 60+). Compared with previous generations, those aged 60+ have fewer children, on average, and may only be able to rely on one child for support in later life. These economic and demographic shifts challenge the traditional notion and value of intergenerational living arrangements for young and old alike. These changes may have significant implications for older adults’ SWB and thus population health.

In response to an aging society, a growing stream of research is exploring how living arrangements relate to older adults’ SWB. However, the categorization of living arrangements used in previous studies may not fully reflect the diversity of household structures and thus their association with SWB. Consequently, much remains to be learned. This paper contributes to this line of research by using a more refined classification of living arrangements than prior studies. We also consider how two theoretically important contextual factors—income and social position—might influence the association between living arrangements and SWB (Ferring & Boll, 2010).

Older Adults’ Subjective Well-Being

A rich body of literature explores how to define and measure SWB (Ferring & Boll, 2010; George, 2010). Generally, subjective well-being is understood to either include two components related to cognition and affection (Ferring & Boll, 2010) or to encompass evaluative, eudemonic, and affective components (Dolan & Metcalfe, 2012). Life satisfaction (e.g., “how satisfied you are with your life overall?”), a global appraisal of one’s life, is an example of a cognitive (or evaluative) component of SWB (Diener, 1994). Happiness is an example of the affective component of SWB (Chen, 2019). How much control one feels is an example of a eudemonic approach (Ryff & Singer, 1998). In this study, we assess all of these theoretical dimensions of SWB to provide a broad view of older adults’ SWB.

The extant literature has built upon several theoretical precepts to identify contributing factors to SWB, including discrepancy theory, social comparisons, the strategic investment of resources, and social stratification theory (see George, 2010, for details). These theories suggest that the determinants of SWB are related to individual and contextual factors such as age, gender, physical health, social position (e.g., socioeconomic status), current life circumstances (e.g., social relationships), critical life events (e.g., disability, widowhood), and intentional activities (e.g., social integration and engagement) (Ferring & Boll, 2010; George, 2010). Specifically, discrepancy theory emphasizes that SWB tends to be lower when there is a discrepancy between aspirations and reality. Social comparison theory emphasizes that one’s subjective perception depends on how one is doing compared to others, such as how older adults perceive themselves relative to younger generations. When older adults strategically invest their resources, they may increase their SWB by increasing their emotional investment in more intimate or rewarding relationships and letting go of less important ones. Social stratification theory underscores the importance of social structures and processes that determine the allocation of SWB-promoting resources. Societal markers like high levels of education or high income come with more resources that are conducive to maximizing SWB.

Building upon these theories, research has consistently found that lower SWB is related to having chronic illnesses or functional disabilities (George, 2010) and that social position as characterized by income is a strong predictor of SWB (Ferring & Boll, 2010; George, 2010). The role of personal relationships in determining one’s SWB is also widely documented. For example, being married is significantly related to higher SWB (George, 2010). The association between relationships with adult children and SWB, however, is less clear cut. Some studies have found positive associations with SWB (George, 2010), whereas others have found the opposite, citing intergenerational conflict and a lack of privacy when living with children and/or grandchildren (Chen & Short, 2008). Previous studies have also found that living with others, like friends, may promote SWB due to perceived social support: the more perceived support, the higher the SWB (George, 2010; Pinquart & Sorensen, 2000). Similarly, being an active member of the community, such as by volunteering, is a form of social engagement found to promote SWB (George, 2010).

The Case for China

While our study was guided by theories and empirical studies on the determinants of SWB using both Western and non-Western samples (Ferring & Boll, 2010; George, 2010), we also considered how cultural traditions and recent macro-level transformations in China could help shape our analysis. The cultural and structural changes currently transforming modern China have likely impacted familial norms and household arrangements. Traditional Chinese culture endorses filial piety and the belief that children, typically the eldest son, should be the primary caregivers for their aging parents (Chow, 2009). Older adults in China might thus expect to live with their children in later life; if this does not occur, the discrepancy between aspirations and reality would theoretically predict lower SWB (George, 2010).

The body of research related to older adults’ SWB in China is growing (e.g., Chen, 2019; Chen & Short, 2008; Chen & Silverstein, 2000; Silverstein et al., 2006; Zhang, 2015), with many studies finding that living arrangements matter and that living with children, either with or without grandchildren, promotes older adults’ SWB (e.g., Chen, 2019; Chen & Short, 2008). One exception was a recent study using data from the capital city, Beijing. The authors found that those living with only one’s spouse or in three-generation households had better SWB than those living in two-generation households (Ye et al., 2017). Generally, living alone appears to be the worst living arrangement for promoting SWB, likely due to discrepancies in what older adults would like to have (living with children or others) versus reality (living alone) (Chen & Short, 2008; Zhou et al., 2015).

Growing evidence suggests, however, that these norms may be changing as a product of the one-child policy and China’s massive internal migration. China’s one-child policy, implemented in 1979 to control population growth, has compromised the two- or three-generation household norm as contemporary Chinese older adults often only have one child to rely on for support (Zhu, 2003). To compound this issue, a massive flow of young adults migrating from rural to urban areas has left many aging parents behind in rural communities (Chen & Silverstein, 2000). China’s 2010 national census showed that the percentage of adults aged 65+ living with children had decreased by 52% since the previous census in 2000, and those living alone or with just their spouse had increased by 42% (Wang, 2014). The dominant household arrangements for older adults as of 2010 were spousal dyads and three-generation households, each accounting for around 30% of older adults (Wang, 2014). Thus, compared to previous generations, current and future generations of older adults are less likely to have adult children as a primary source of financial and emotional support. Empirical studies have confirmed that many older adults in contemporary China have extended their support networks in response (e.g., Ng et al., 2002). Evidence also indicates that preferences in living arrangements may be shifting, with more older adults desiring to live independently (e.g., with a spouse) when financially and physically feasible (Tang & Yu, 2016). The expansion of the welfare state in urban areas may also enable greater independence by improving older adults’ overall physical health and economic security (Tang & Yu, 2016). Similarly, recent improvements in professional elder care services and facilities, particularly in urban areas, has increased access to formal support (Chen & Han, 2016).

The SWB literature also suggests that social stratification is linked to SWB among older adults (George, 2010). Stratification theory emphasizes that social structures create differential allocations of resources and assets to members of society; thus, SWB is expected to be highest among those with the most resources, such as those with high income (Ferring & Boll, 2010; George, 2010). We include in our analysis a factor that uniquely represents social stratification in China: the household registration system or hukou. The hukou system documents one’s place of birth and distinguishes between those with urban and rural origins as though they are of different social castes. A rural hukou usually comes with limited access to healthcare and government entitlements, whereas an urban hukou correlates with higher levels of education, social status, and family income (Wu & Zheng, 2018). Hukou status thus not only represents socioeconomic status but also reflects cultural and institutional differences in China. In this study, we considered both hukou and income as social stratification factors that could potentially shape the association between living arrangements and SWB.

We note, however, that our understanding of this association in China was built upon studies that used varying categorizations of living arrangements. For example, using data from the Chinese Longitudinal Healthy Longevity Survey (CLHLS), Zhang (2015) used three categories of living arrangements: living alone, with household members, or in an institution. Chen and Short (2008) also used the CLHLS but provided eight categories of living arrangements: living with a spouse, no children; living with children (son or daughter), no spouse; living with a spouse and children (son or daughter); and living with no spouse, no children. Chen (2019), using data from the China Health and Retirement Longitudinal Study (CHARLS), identified five types of living arrangements: living with adult children, adult children living nearby, adult children living far away, living in a nursing home, or other arrangements.

In comparison, we took advantage of the data available on household rosters to create more nuanced categories of living arrangements: living with a spouse only, in two-generation households, in three-generation households, in skipped-generation households (i.e., with grandchildren but not children), living alone, and “other” living arrangements (e.g., living with non-children adults). We also differentiated whether the older adult was living with a son/daughter or a son/daughter-in-law (Cong & Silverstein, 2008).

Building upon theories, empirical evidence, and Chinese cultural norms, we hypothesized that living with a spouse or with adult children in two- or three-generation households (without or with grandchildren) would be associated with higher SWB than those in other living arrangements. The theory of strategic investment of resources suggests that such living arrangements promote SWB by increasing the emotional investment of older adults in more intimate or rewarding relationships with family members. The comparison between living with a spouse and living in two- or three-generation households for a better SWB, however, would be an empirical question. Similarly, living only with grandchildren (skipped-generation) would likely increase SWB. However, such households may also be more economically vulnerable as they are a product of young adult children leaving home for better employment and financial opportunities.

Based on social stratification theory, which emphasizes the socioeconomic context, we further hypothesized that the positive association between skipped-generation arrangements and SWB might depend on family resources. Likewise, resources are likely to influence SWB in two- or three-generation households; lack of resources might weaken the positive association between these living arrangements and SWB. In addition to income, an objective indicator of resources, we considered hukou status, which in China determines one’s legal residence and thus access to welfare entitlements (e.g., pension) and resources. The hukou system thus constitutes not only a socioeconomic context but also an institutional context for unequal access to resources, which could potentially create disparities in well-being between rural and urban residents.

Methods

Data

We utilized data collected in Wave 1 of the World Health Organization’s Study on Global Ageing and Adult Health (SAGE), a nationally representative longitudinal dataset administered in six countries: China, Ghana, India, Mexico, Russia, and South Africa. Wave 1 was collected between 2007 and 2010 and included a total sample of 14,813 respondents in China over the age of 18. Wave 2 of SAGE (collected in 2014/15) is not yet publicly available (https://www.who.int/healthinfo/sage/cohorts/en/).

The SAGE survey contains measurements used in similar country-administered longitudinal studies on aging, like the American Health and Retirement Study, and instruments from the WHO World Health Survey, which is administered in 70 countries. The unit of analysis for this study was individual respondents over the age of 50, yielding a final sample size of 13,367. SAGE instruments assess 11 main domains of respondents’ lives: sociodemographic characteristics, work history and insurance/retirement benefits, health status, physical and biomarker measures, risk and preventative health behaviors, chronic health conditions, health services coverage, health care utilization, social cohesion, subjective well-being and quality of life, and impact of caregiving.

Measures

Table 1 presents the details measuring each of the dependent and independent variables used in this analysis.

Table 1.

Measures for All Analyzed Variables.

Variables	Definition	Values/Categories	Cronbach’s Alpha (α)
Dependent Variables
Subjective Well-Being (SWB)	Measured by four variables
Global SWB	12 items on a scale of 1–5 (very satisfied to very dissatisfied): having enough energy for everyday life, having enough money to meet needs, satisfied with health, satisfied with themselves, satisfied with their ability to perform activities of daily life (ADLs), satisfied with personal relationships, satisfied with living conditions, satisfied with life as a whole, ability to control important things in their lives, how often they were unable to cope with stressors, the overall rating of their quality of life, and their level of overall happiness.	A standardized variable with a mean of 0 and a standard deviation of 1. Higher score represents higher SWB.	0.90
Life Satisfaction	One item asked respondents about their satisfaction with life as a whole on a scale of 1 (very satisfied) to 5 (very dissatisfied).	A standardized variable with a mean of 0 and a standard deviation of 1. Higher score represents higher life satisfaction.	NA
Happiness	One item asked respondents about their overall happiness on a scale of 1 (very happy) to 5 (very unhappy).	A standardized variable with a mean of 0 and a standard deviation of 1. Higher score represents higher happiness levels.	NA
Control	Two items measuring how much controls one feels: the ability to control important things in their life and how often they were unable to cope with stressors. Responses ranged from 1(never) to 5 (very often).	A standardized variable with a mean of 0 and a standard deviation of 1. Higher score represents higher levels of perceived control.	0.91
Independent Variables
Living Arrangement	Six categories created from asking respondents who they were currently living with by recording their age, gender, marital status, and relationship with the respondent: Living with spouse only, two-generation (i.e., living with children only, with or without spouse), three-generation (i.e., living with both children and grandchildren, with or without spouse), skipped-generation (i.e., living with grandchildren only, with or without spouse), living alone, or all other arrangements.	A set of dichotomous variables with a 0/1 value for each of the six categories.
Living with daughter	Identified from household roster if the respondent currently was living with a daughter.	A dichotomous variable with 0/1 value.
Living with daughter-in-law	Identified from household roster if the respondent currently was living with a daughter-in-law.	A dichotomous variable with 0/1 value.
Living with son	Identified from household roster if the respondent currently was living with a son.	A dichotomous variable with 0/1 value.
Living with son-in-law	Identified from household roster if the respondent currently was living with a son-in-law.	A dichotomous variable with 0/1 value.
Physical Health	Measured with three indicators.
Self-rated health	Asked respondents to rate their health as very good, good, moderate, bad, or very bad.	A dichotomous variable indicating whether the respondent reported having very bad or bad health (1), and otherwise (0).
Functional health	Measured by 44 items indicating the degree of difficulty individuals had with ADLs and instrumental ADLs.	A standardized score with a mean of 0 and a standard deviation of 1. Higher score represents better functional health.	0.97
Chronic health conditions	Measured by nine dichotomous items indicating whether the respondent had ever been diagnosed with arthritis, a stroke, angina, diabetes, chronic lung disease, asthma, depression, or cataracts or had lost all their teeth.	A set of dummy variables based on the sample distribution to indicate if the respondents had none of these conditions (the reference group), had one chronic condition, or had two, or had three or more.
Social Engagement	Measured using nine items to capture the frequency with which older adults interacted with social groups or friends over the previous 12 months. These items included attending public meetings, meeting with a community leader, attending a club or group meeting, working with people in their neighborhood, interacting with friends at their home, interacting with friends at the respondent’s home, socializing with colleagues, attending a religious service, and going out in their neighborhood. Each of these items was measured with five response categories: never, 1 or 2 times per year, 1 or 2 times per month, 1 or 2 times per week, or daily.	We coded each of these nine items to create a set of dichotomous variables indicating whether the respondent had participated in the activity at least once in the last 12 months (1) or not (0). We then summed and standardized the score with a mean of 0 and a standard deviation of 1. Higher score represents the respondent was more engaged.	0.66
Sociodemographic background	Following prior research on the determinants of SWB, we included in our analysis a robust set of sociodemographic characteristics shown to be important to older adults’ SWB in China (Chen & Short, 2008; George, 2010). Including such characteristics also helps address selection bias.
Age	Recorded at the time of survey completion.	A continuous variable
Gender	Reported dichotomously and coded as female (0) or male (1).	A dichotomous variable (0/1 value)
Education	Less than primary school (<6 years), primary school (=6 years, reference group), middle school (7–9 years), high school (10–12 years), or college or higher (>12 years).	Five dichotomous variables with a value of 0/1.
Marital status	A categorical variable is presented descriptively in Table 2 with the following groups: married, widowed, divorced or separated, or not married. Due to small sample sizes for the unmarried category, we used a dichotomous variable (not married vs. married as the reference group) in the multivariate regression analyses. This change yielded no significant differences in our regression estimates.	A dichotomous variable with a value of 1 for not married and 0 for married.
Number of people in HH	Total number of people living in a household.	A continuous variable
Hukou status	Respondent’s birthplace.	A dichotomous variable with a value of 1 for rural and 0 otherwise.
Family income	We calculated household income over the past 12 months by dividing the total household income by the number of household members. We then transformed income into a log value to correct for skewness.	A continuous variable (log value)
Employment status	Using 11 categories that include caregiving roles, homemaking, full-time schooling, volunteer work, and other types of disengagement from the labor force. Most respondents reported working full-time (38.84%) or retirement/not working (48.91%) as their primary work status. As such, we recoded these 11 categories into a dichotomous variable. Almost 80% of respondents over the age of 65 had left the paid workforce.	A dichotomous variable with a value of 1 for working and 0 for otherwise.
Risky health behaviors	Using two dichotomous items indicating whether individuals had ever smoked cigarettes or had ever drunk alcohol in their lifetime.	A set of dichotomous variables indicating if an individual had never smoked or drunk (reference group), ever smoked or drunk, or ever smoked and drunk.

Empirical Strategy

The response rate for all survey items was higher than 95%. About 4% had missing values on income and social engagement items. Analysis of these missing values indicated that those with missing income responses compared to those without were on average older (67.8 vs. 61.5 years old) and were more likely to be female (71.8% vs. 46.3%) and to hold rural hukou status (70.7% vs. 43.7%). Little’s (1988) MCAR test indicated that these missing values on income were not missing completely at random but that across all covariates there was no overall pattern to missingness. We thus conducted multiple imputations for income and social engagement variables for those missing cases by using chained equations in STATA (White et al., 2011).

We used multivariate regression analysis to estimate the relationship between living arrangements and older adults’ SWB, controlling for all sociodemographic characteristics discussed above. We applied sampling weights in our regression estimates to produce Huber-White standard errors that are robust to unspecified heteroskedasticity (Greene, 2012). While these ordinary least squares regression estimates cannot be used to infer causality given the cross-sectional nature of the data, these estimates can be used to inform variables of interest for future longitudinal analyses of WHO SAGE data. We conducted separate analyses by age for those aged 50+ and those aged 65+. China’s retirement age was 60 for men, 50 for women workers in enterprises, and 55 for women civil servants during our sample period. We conducted separate analyses for those aged 65+ to be comparable to other studies focused on other countries. Because many of our variables are standardized, the coefficients represent the standardized mean difference in the dependent variable with a one standard deviation change in the independent variable. Thus, the results can be interpreted as effect sizes (Nieminen et al., 2013).

We next used interaction analyses to investigate theoretically informed moderating effects of social stratification variables on the association between living arrangements and SWB. Specifically, we divided household income into terciles. We treated hukou status as a dichotomous variable. The interaction analyses used the dichotomous indicators interacted with the living arrangement variable, again separately for those aged 50+ and those 65+. For example, interacting hukou status with six types of living arrangements produced 12 mutually exclusive groups. Individuals with urban hukou and living with a spouse only was the reference group. Similarly, we interacted family income terciles with the six living arrangements, producing 18 mutually exclusive groups. Individuals in the bottom third of the household income distribution who were living with a spouse only was the reference group.

Results

Descriptive Picture

Table 2 presents descriptive data on all analyzed variables for the total sample as well as separately by living arrangements. Overall, much of the sample (41%) lived with a spouse only, followed by those living in two-generation households (25%), in three-generation households (14%), and then those living alone (12%). Another 4% lived in skipped-generation households, and 4% had other living arrangements. The most common form of “other arrangement” was co-residing with a sibling or other unidentified family member. About 20% of the respondents lived with their daughters or their daughters-in-law and about 11% lived with their sons or sons-in-law. The average age of our respondents was about 63 years old (SD = 9.44), over half were female (54.06%), more than half had low levels of education (63% completed primary school or less), the majority were married (83%), and only slightly over one-third were working. The average annual household income was about RMB 27,000, equivalent to US $4,500. About 20% of respondents rated their health as bad or very bad. Over half of respondents reported having at least one chronic health condition, but less than half reported ever smoking or drinking in their lifetime.

Table 2.

Descriptive Statistics of All Analyzed Variables, WHO SAGE Wave 1.

		Living Arrangements
Variables	Total Sample	W/ Spouse Only	Two Generation	Three Generation	Skipped Generation	Living Alone	All Other Arrangements
N	13,367	5,481	3,379	1,868	522	1,594	523
Age ***	63.16 (9.44)	63.52 (8.86)	60.16 (9.17)	63.46 (8.67)	61.93 (7.59)	67.19 (10.48)	66.53 (11.39)
Male (%) ***	46.94	50.87	47.88	45.29	44.64	36.26	40.34
Rural birthplace (Hukou) (%) ***	50.89	53.69	39.23	53.83	70.69	51.02	66.22
Highest Education Level (%) ***
Less Than Primary (< 6 yrs)	44.14	40.62	37.07	49.76	51.72	56.99	59.16
Primary (6 yrs)	19.35	19.92	18.65	22.85	21.07	15.71	15.08
Secondary (7–9 yrs)	19.47	19.65	23.77	17.28	17.43	14.36	15.65
High School (10–12 yrs)	12.56	13.40	16.58	7.97	7.85	9.86	7.25
College or Plus (13+ yrs)	4.47	6.42	3.93	2.14	1.92	3.08	2.86
Marital Status (%) ***
Married	82.81	99.22	83.69	80.94	87.36	33.81	56.70
Widowed	14.25	0.26	13.76	16.97	10.73	55.02	33.72
Divorced/Separated	1.67	0.09	1.86	1.28	1.34	6.71	3.26
Other	1.26	0.44	0.68	0.80	0.57	4.45	6.32
# of Household Members ***	2.75 (1.31)	2.01 (0.17)	3.37 (0.94)	4.91 (0.98)	3.05 (0.73)	1.28 (0.86)	2.90 (0.74)
Currently Working (%) ***	38.84	40.18	38.20	36.04	55.84	33.29	39.36
Household Income In RMB Last Year ***	27061.50 (47257.86)	23812.52 (38687.26)	36168.98 (66212.37)	36461.55 (47400.45)	16237.63 (20769.34)	11953.97 (16785.57)	24753.64 (44167.08)
Log Household Income Last Year ***	8.66 (1.14)	8.74 (1.20)	8.78 (1.07)	8.56 (1.02)	8.06 (1.13)	8.53 (1.15)	8.48 (1.06)
Social Engagement ***	0.00 (0.52)	0.003 (0.50)	0.002 (0.53)	0.02 (0.54)	0.04 (0.47)	−0.03 (0.53)	−0.07 (0.54)
Self-Rated Health Poor (%) ***	20.55	20.22	18.70	19.06	22.18	25.59	24.15
Functional Health ***	−0.00 (0.66)	0.00 (0.65)	0.06 (0.62)	0.03 (0.62)	0.05 (0.50)	−0.14 (0.74)	−0.17 (0.89)
Chronic Health Issues (%) ***
None	45.05	43.99	48.12	43.61	51.34	40.73	48.47
1	28.64	28.61	28.76	29.80	27.01	28.34	26.53
2	14.83	15.36	13.86	14.02	13.03	16.64	14.69
3+	11.48	12.03	9.25	12.57	8.62	14.29	10.31
Risky Health Behaviors (%) *
Never Smoked or Drank Alcohol	56.55	56.51	55.45	55.04	50.20	62.97	55.65
Either Smoked or Drank	22.71	22.63	23.38	22.70	25.30	19.85	25.80
Have Smoked and Drank	20.74	20.86	21.17	22.26	24.51	17.18	18.55
Living w/ Daughter (%) ***	21.92	na	37.84	11.83	na	na	na
Living w/ Daughter-in-Law (%) ***	19.22	na	17.68	37.94	na	na	na
Living w/ Son (%) ***	10.83	na	40.23	4.87	na	na	na
Living w/ Son-in-Law (%) ***	1.19	na	2.39	4.23	na	na	na
Outcomes
Global SWB ***	0.00 (0.70)	0.03 (0.69)	−0.01 (0.70)	0.06 (0.70)	0.01 (0.62)	−0.12 (0.71)	−0.10 (0.72)
Life Satisfaction ***	0.00 (1.00)	0.06 (0.98)	−0.05 (1.02)	0.05 (0.99)	0.02 (0.93)	−0.13 (1.03)	−0.08 (0.99)
Happiness ***	0.00 (1.00)	0.04 (0.96)	−0.05 (1.01)	0.05 (1.05)	−0.00 (0.99)	−0.07 (1.03)	−0.11 (0.98)
Total Control ***	0.00 (1.00)	−0.01 (0.97)	0.04 (1.05)	0.13 (0.98)	−0.00 (0.88)	−0.10 (0.92)	−0.29 (1.23)

Note. Numbers represent either means (standard deviations in parentheses) or percentages. Bivariate statistical significant tests were performed by χ² for categorical variables or by ANOVA for continuous variables.

*p < .05, **p < .01, ***p < .001.

We see notable differences in these sociodemographic characteristics across living arrangements. For example, adults aged 50+ in the sample who lived either alone or in all “other” arrangements had the most disadvantaged characteristics relative to those living in multigeneration households with or without their spouses. These two groups, living alone or in other arrangements, were significantly older and were more likely to be female, widowed, and have lower educational levels (almost 60% had less than 6 years of schooling). They were also more likely to rate their health as poor and had worse functional health, but they were more likely to report never smoking or drinking. It is no surprise given this descriptive picture that those living alone or in other arrangements had the lowest SWB in the areas of life satisfaction, happiness, and control. In contrast, those living with a spouse only or in two- or three-generation households tended to have relatively advantaged sociodemographic characteristics. For example, they were more likely to be male, to be married, to have higher household income, and to be more socially engaged. In addition, they were less likely to rate their health as poor and tended to have better functional health than their counterparts. Also, those living in skipped-generation households were relatively younger and were more likely to be female with low education levels and low income. The two groups with the lowest household income in absolute dollar amount were those living alone and those in skipped-generation households.

Not surprisingly, groups with relatively more advantaged backgrounds tended to have better SWB, but the significant differences in these sociodemographic characteristics among the various living arrangement groups also reveals a selection story. We next used multivariate regression analysis to estimate the net association between living arrangements and SWB by considering these sociodemographic characteristics.

Regression Results

Tables 3-1 and 3-2 present our main analysis of the net association between living arrangements and older adults’ SWB outcomes for the 50+ and 65+ samples, respectively. We controlled for all sociodemographic characteristics as well as for physical health, social engagement, and risky health behaviors. The sample sizes for each of the four outcomes are slightly different due to missing values. We followed previous literature and did not impute the missing values for the outcome variables to avoid noise in the measurements (von Hippel, 2007).

Table 3-1.

Regression Estimates of the Association Between Living Arrangements and Subjective Well-Being (50+).

	Global SWB		Life Satisfaction		Happiness		Control
	b	se	b	se	b	se	B	se
Two Generations	−0.10^***	(0.02)	−0.15^***	(0.04)	−0.17^***	(0.04)	0.02	(0.03)
Three Generations	−0.03	(0.03)	−0.09	(0.05)	−0.16^**	(0.06)	0.07	(0.04)
Skip Generations	0.03	(0.03)	0.00	(0.05)	−0.05	(0.05)	0.08^*	(0.04)
Living Alone	−0.02	(0.02)	−0.08^*	(0.04)	0.00	(0.04)	0.03	(0.03)
All Other Living Arrangements	−0.01	(0.04)	−0.04	(0.06)	−0.14^*	(0.06)	0.03	(0.04)
Lives with Daughter	0.08^***	(0.02)	0.06	(0.04)	0.10^**	(0.04)	0.03	(0.02)
Lives with Daughter-in-Law	0.04	(0.02)	0.05	(0.04)	−0.01	(0.04)	0.11^***	(0.03)
Lives with Son	−0.00	(0.03)	−0.02	(0.05)	−0.02	(0.05)	−0.01	(0.04)
Lives with Son-in-Law	−0.00	(0.05)	−0.04	(0.09)	−0.09	(0.09)	0.05	(0.08)
Male	−0.01	(0.02)	−0.04	(0.02)	−0.03	(0.03)	0.04	(0.03)
Age	0.01^***	(0.00)	0.02^***	(0.00)	0.02^***	(0.00)	0.00^**	(0.00)
Not Married	−0.08^***	(0.02)	−0.06	(0.03)	−0.07^*	(0.03)	−0.10^***	(0.03)
Education
< Primary (< 6 years)	−0.06^***	(0.02)	−0.08^**	(0.03)	−0.09^**	(0.03)	−0.02	(0.02)
Middle School (7–9 years)	0.02	(0.02)	0.03	(0.03)	−0.02	(0.03)	−0.01	(0.02)
High School (10–12 years)	−0.01	(0.02)	0.01	(0.03)	0.01	(0.03)	−0.11^***	(0.03)
College + (13+ years)	0.02	(0.03)	0.02	(0.05)	−0.05	(0.05)	−0.09^*	(0.04)
Rural Hukou Status	0.04^*	(0.02)	0.14^***	(0.03)	0.03	(0.03)	−0.06^*	(0.02)
# of Household Members	0.01	(0.01)	0.02	(0.01)	0.06^***	(0.02)	0.02	(0.01)
Currently Working	0.09^***	(0.01)	0.12^***	(0.02)	0.10^***	(0.03)	0.05^**	(0.02)
Log of Income Per Capita	0.11^***	(0.01)	0.14^***	(0.01)	0.10^***	(0.01)	0.06^***	(0.01)
Social Engagement	0.06^***	(0.01)	0.06^**	(0.02)	0.11^***	(0.02)	0.02	(0.02)
Self-Rated Health Poor	−0.35^***	(0.02)	−0.36^***	(0.03)	−0.33^***	(0.03)	−0.15^***	(0.02)
Functional Health	0.40^***	(0.01)	0.44^***	(0.02)	0.36^***	(0.02)	0.31^***	(0.02)
Chronic Health Condition
1 Chronic Condition	−0.08^***	(0.01)	−0.08^***	(0.02)	−0.03	(0.02)	−0.03	(0.02)
2 Chronic Conditions	−0.13^***	(0.02)	−0.12^***	(0.03)	−0.06^*	(0.03)	−0.05	(0.02)
3+ Chronic Conditions	−0.16^***	(0.02)	−0.20^***	(0.04)	−0.10^**	(0.04)	−0.03	(0.03)
Risky Health Behaviors
Smoked or Drank	−0.01	(0.02)	−0.03	(0.03)	−0.00	(0.03)	0.02	(0.02)
Smoked and Drank	−0.00	(0.02)	−0.03	(0.03)	−0.03	(0.03)	0.04	(0.03)
_cons	−1.48^***	(0.10)	−2.09^***	(0.15)	−1.84^***	(0.15)	−0.65^***	(0.11)
Observations	12,674		12,655		12,629		12,836

Note. Standard errors in parentheses. The reference group is living with spouse only, not living with son/daughters nor son/daughters-in-law, female, with average age, married, having 6 years of education, urban hukou status, average number of household members, not working, average log income, average level of social engagement, self-rated health not being poor, average functional health, having no chronic health condition, and never smoked nor drank.

*p < .05, **p < .01, ***p < .001.

Table 3-2.

Regression Estimates of the Association between Living Arrangements and Subjective Well-Being (65+).

	Global SWB		Life Satisfaction		Happiness		Control
	b	se	b	se	b	se	b	se
Two Generations	−0.17^***	(0.04)	−0.28^***	(0.06)	−0.19^**	(0.07)	−0.06	(0.06)
Three Generations	−0.09^*	(0.05)	−0.21^**	(0.07)	−0.07	(0.09)	−0.04	(0.08)
Skip Generations	−0.05	(0.05)	−0.07	(0.09)	−0.11	(0.11)	0.02	(0.07)
Living Alone	−0.06	(0.03)	−0.14^**	(0.05)	−0.07	(0.06)	−0.04	(0.05)
All Other Living Arrangements	−0.09	(0.06)	−0.15	(0.08)	−0.19^*	(0.09)	−0.07	(0.07)
Lives with Daughter	0.18^***	(0.04)	0.20^**	(0.07)	0.17^*	(0.07)	0.03	(0.03)
Lives with Daughter-in-Law	0.06	(0.04)	0.15^*	(0.06)	−0.04	(0.07)	0.15^*	(0.06)
Lives with Son	0.03	(0.06)	−0.01	(0.10)	0.07	(0.10)	0.16	(0.09)
Lives with Son-in-Law	−0.08	(0.10)	−0.08	(0.15)	−0.18	(0.16)	0.12	(0.15)
Male	−0.02	(0.02)	−0.09^*	(0.04)	−0.00	(0.04)	0.02	(0.06)
Age	0.01^***	(0.00)	0.02^***	(0.00)	0.02^***	(0.00)	0.00	(0.00)
Not Married	−0.04	(0.03)	0.01	(0.05)	−0.03	(0.05)	−0.05	(0.04)
Education
< Primary (< 6 Years)	−0.02	(0.03)	−0.07	(0.05)	−0.07	(0.05)	0.05	(0.04)
Middle School (7–9 Years)	0.07^*	(0.03)	0.10^*	(0.05)	0.03	(0.06)	0.04	(0.04)
High School (10–12 Years)	0.03	(0.04)	0.03	(0.05)	0.04	(0.06)	0.00	(0.05)
College + (13+ Years)	0.04	(0.05)	0.05	(0.07)	−0.07	(0.07)	−0.01	(0.06)
Rural Hukou Status	0.02	(0.03)	0.12^**	(0.05)	0.03	(0.05)	−0.12^**	(0.04)
# of Household Members	0.01	(0.01)	0.01	(0.02)	0.02	(0.02)	0.02	(0.02)
Currently Working	0.08^**	(0.03)	0.10^*	(0.04)	0.02	(0.05)	0.09^**	(0.03)
Log of Income Per Capita	0.10^***	(0.01)	0.12^***	(0.02)	0.06^**	(0.02)	0.06^***	(0.01)
Social Engagement	0.09^***	(0.02)	0.10^***	(0.03)	0.17^***	(0.03)	0.03	(0.02)
Self-Rated Health Poor	−0.38^***	(0.03)	−0.34^***	(0.04)	−0.36^***	(0.04)	−0.18^***	(0.03)
Functional Health	0.35^***	(0.02)	0.39^***	(0.03)	0.32^***	(0.03)	0.27^***	(0.02)
Chronic Health
1 Chronic Condition	−0.05^*	(0.02)	−0.05	(0.04)	0.00	(0.04)	−0.03	(0.03)
2 Chronic Conditions	−0.09^***	(0.03)	−0.11^**	(0.04)	−0.05	(0.05)	−0.00	(0.03)
3+ Chronic Conditions	−0.13^***	(0.03)	−0.20^***	(0.05)	−0.04	(0.05)	0.01	(0.04)
Risky Health Behaviors
Smoked or Drank	0.02	(0.02)	0.02	(0.04)	0.02	(0.04)	0.04	(0.03)
Smoked and Drank	−0.02	(0.03)	0.01	(0.05)	−0.08	(0.05)	0.03	(0.04)
_Cons	−1.48^***	(0.18)	−1.97^***	(0.27)	−1.72^***	(0.27)	−0.67^**	(0.21)
Observations	5,100		5,093		5,082		5,154

Note. Standard errors in parentheses. The reference group is living with spouse only, not living with son/daughters nor son/daughters-in-law, female, with average age, married, having 6 years of education, urban hukou status, average number of household members, not working, average log income, average level of social engagement, self-rated health not being poor, average functional health, having no chronic health condition, and never smoked nor drank.

*p < .05, **p < .01, ***p < .001.

Results in Table 3-1, which focuses on respondents aged 50+, indicate that those living in two-generation households had significantly lower global SWB, lower life satisfaction, and lower happiness than those living with a spouse only. Those in three-generation households reported having significantly lower happiness than those living with a spouse only. The only group that reported having more control over their lives than those living with only a spouse were respondents living with grandchildren only (i.e., skipped-generation households). As expected, people living alone reported significantly lower life satisfaction, and those with “other” living arrangements reported significantly lower happiness than respondents living with a spouse only. Notably, older adults living with their daughters reported significantly higher global SWB and happiness than those living without. And, living with daughters-in-law versus without was associated with feeling more control among those aged 50+.

Results in Table 3-2, using the 65+ sample, suggest similar patterns to those in Table 3-1 but with stronger associations. For example, adults aged 65+ living in two- or three-generation households reported significantly lower global SWB and life satisfaction than their counterparts who lived with a spouse only, and those living in two-generation households also reported significantly lower happiness. However, among adults 65+, living in skipped-generation households was not associated with the same feeling of control as their younger counterparts (see Table 3-1).

The associations between sociodemographic characteristics and older adults’ SWB are consistent with previous findings (e.g., Ferring & Boll, 2010; George, 2010). Briefly, being older and married, having a high income, working, and having better physical health were all associated with significantly higher SWB than their corresponding counterparts. Reports of low educational attainment and poor self-rated health were associated with significantly lower SWB. Having a rural hukou, in general, was associated with better global SWB and life satisfaction, but also with significantly lower perceived control.

Did Social Stratification Modify the Association Between Living Arrangements and SWB?

Results in Tables 3-1 and 3-2 paint a picture consistent with prior studies yet reveals complexities in how older adults felt on different domains of SWB across different living arrangements. Social stratification theory predicts that more resources promote positive SWB and can cushion adverse effects associated with life circumstances (George, 2010). Of note, the two living arrangement groups with the highest household income in our samples were those living in two- or three-generation households; however, results in Tables 3-1 and 3-2 do not necessarily tell a straightforward story. We next present interaction analyses by income and by hukou, two indicators representing social stratification in China.

For brevity, we present only the interaction terms in Tables 4 and 5, but these analyses controlled for all variables in Tables 3-1 and 3-2. Table 4 presents the interaction analysis by income on four outcome variables. Panels A and B show results for the 50+ and 65+ samples, respectively. Table 5 presents results on the interaction analysis by hukou status and is similarly organized. Results in Table 4 suggest that the adverse association between two-generation households and SWB was more pronounced for those in the bottom third of the household income distribution. Similarly, any adverse associations between three-generation households and SWB were most pronounced among those in the bottom third of the income distribution. Table 4 results also suggest that high income mitigates potential adverse associations between living arrangements and SWB, as all living arrangements in the high-income group reported significantly higher SWB in all domains than those living with a spouse only and having income in the bottom third of the distribution. Indeed, those with mid-third income who were living with a spouse only also had significantly higher SWB in most domains than their counterparts with income in the bottom third. Similar patterns of results were found when we restricted our sample to those aged 65+ (Panel B of Table 4); however, the cushioning effect associated with income was not as obvious among this older age group. For example, people aged 65+ living in either two-generation or skipped-generation households with income in the top third did not have significantly higher life satisfaction or happiness than those in the bottom third living with a spouse only.

Table 4.

Regression Estimates of the Association Between Living Arrangements and Subjective Well-Being by Income.

	Global SWB		Life Satisfaction		Happiness		Control
	b	se	b	se	b	se	b	se
Panel B. Among 65+
Income = 1 & with Spouse Only	Reference	Group	Reference	Group	Reference	Group	Reference	Group
Income = 1 & Two Generation	−0.16^***	(0.03)	−0.24^***	(0.05)	−0.29^***	(0.06)	−0.07	(0.04)
Income = 1 & Three Generation	−0.07	(0.04)	−0.17^*	(0.07)	−0.24^**	(0.07)	0.03	(0.05)
Income = 1 & Skip Generation	0.01	(0.05)	−0.02	(0.07)	−0.08	(0.07)	0.02	(0.05)
Income = 1 & Living Alone	−0.05	(0.04)	−0.12^*	(0.06)	−0.04	(0.06)	0.01	(0.04)
Income = 1 & All Other Arrangements	0.04	(0.06)	0.13	(0.09)	−0.21^*	(0.10)	−0.04	(0.06)
Income = 2 & with Spouse Only	0.10^***	(0.03)	0.10^*	(0.04)	0.08	(0.04)	0.10^***	(0.03)
Income = 2 & Two Generation	0.01	(0.03)	−0.00	(0.05)	−0.11^*	(0.05)	0.06	(0.04)
Income = 2 & Three Generation	0.08^*	(0.04)	0.04	(0.06)	−0.10	(0.07)	0.09	(0.05)
Income = 2 & Skip Generation	0.17^***	(0.05)	0.16^*	(0.08)	0.07	(0.09)	0.19^***	(0.05)
Income = 2 & Living Alone	0.09^*	(0.04)	0.07	(0.06)	0.03	(0.07)	0.09^*	(0.04)
Income = 2 & All Other Arrangements	0.07	(0.06)	0.03	(0.09)	−0.08	(0.09)	0.03	(0.06)
Income = 3 & with Spouse Only	0.30^***	(0.03)	0.39^***	(0.04)	0.25^***	(0.04)	0.07^*	(0.03)
Income = 3 & Two Generation	0.26^***	(0.03)	0.30^***	(0.06)	0.17^**	(0.06)	0.21^***	(0.05)
Income = 3 & Three Generation	0.31^***	(0.05)	0.36^***	(0.07)	0.18^*	(0.09)	0.29^***	(0.07)
Income = 3 & Skip Generation	0.26^**	(0.08)	0.27^*	(0.13)	0.03	(0.13)	0.25^*	(0.13)
Income = 3 & Living Alone	0.27^***	(0.04)	0.32^***	(0.06)	0.31^***	(0.06)	0.13^**	(0.05)
Income = 3 & All Other Arrangements	0.26^**	(0.09)	0.17	(0.12)	0.22	(0.13)	0.30^**	(0.11)
Observations	11964		11945		11919		12126
Panel B. Among 65+
Income = 1 & with Spouse Only	Reference	Group	Reference	Group	Reference	Group	Reference	Group
Income = 1 & Two Generation	−0.18^***	(0.05)	−0.31^***	(0.09)	−0.26^**	(0.10)	−0.14^*	(0.07)
Income = 1 & Three Generation	−0.11	(0.07)	−0.32^**	(0.11)	−0.27^*	(0.12)	−0.10	(0.09)
Income = 1 & Skip Generation	−0.07	(0.07)	−0.02	(0.12)	−0.13	(0.16)	−0.08	(0.09)
Income = 1 & Living Alone	−0.06	(0.05)	−0.12	(0.08)	−0.19^*	(0.08)	−0.00	(0.06)
Income = 1 & All Other Arrangements	0.00	(0.07)	0.15	(0.12)	−0.43^**	(0.15)	−0.06	(0.09)
Income = 2 & with Spouse Only	0.19^***	(0.04)	0.17^*	(0.07)	0.05	(0.07)	0.17^***	(0.04)
Income = 2 & Two Generation	0.01	(0.06)	−0.05	(0.09)	−0.17	(0.10)	0.08	(0.08)
Income = 2 & Three Generation	0.10	(0.06)	0.03	(0.10)	−0.01	(0.12)	0.07	(0.09)
Income = 2 & Skip Generation	0.32^***	(0.08)	0.24	(0.16)	0.05	(0.17)	0.26^*	(0.13)
Income = 2 & Living Alone	0.10	(0.06)	0.07	(0.09)	−0.06	(0.10)	0.08	(0.07)
Income = 2 & All Other Arrangements	0.13	(0.07)	0.03	(0.11)	0.00	(0.11)	−0.01	(0.10)
Income = 3 & with Spouse Only	0.35^***	(0.04)	0.47^***	(0.06)	0.17^*	(0.07)	0.16^**	(0.05)
Income = 3 & Two Generation	0.20^**	(0.07)	0.18	(0.11)	0.04	(0.11)	0.14	(0.10)
Income = 3 & Three Generation	0.30^***	(0.07)	0.30^**	(0.11)	0.28^*	(0.14)	0.26^*	(0.11)
Income = 3 & Skip Generation	0.20	(0.14)	0.26	(0.23)	0.03	(0.22)	0.44^*	(0.22)
Income = 3 & Living Alone	0.25^***	(0.05)	0.29^***	(0.09)	0.19^*	(0.09)	0.10	(0.07)
Income = 3 & All Other Arrangements	0.14	(0.15)	−0.02	(0.16)	0.12	(0.23)	0.17	(0.18)
Observations	4,763		4,756		4,745		4,817

Note. Standard errors in parentheses. Models controlled for all other covariates as shown in Table 2.

*p < .05. **p < .01. ***p < .001.

Table 5.

Regression Estimates of the Association Between Living Arrangements and Subjective Well-Being by Hukou Status.

	Global SWB		Life Satisfaction		Happiness		Control
	b	se	b	se	b	se	b	se
Panel A. Among 50+
Urban & Living with Spouse Only	Reference	Group	Reference	Group	Reference	Group	Reference	Group
Urban & Two Generation	−0.07^**	(0.03)	−0.13^**	(0.04)	−0.10^*	(0.04)	−0.10^*	(0.04)
Urban & Three Generation	−0.00	(0.04)	−0.06	(0.06)	−0.07	(0.07)	−0.07	(0.07)
Urban & Skip Generation	−0.04	(0.05)	−0.19^*	(0.09)	−0.10	(0.10)	−0.11	(0.10)
Urban & Living Alone	−0.01	(0.03)	−0.10^*	(0.04)	0.05	(0.05)	0.05	(0.05)
Urban & All Other Arrangements	0.04	(0.07)	−0.02	(0.09)	−0.09	(0.12)	−0.09	(0.11)
Rural & Living with Spouse Only	0.07^**	(0.02)	0.15^***	(0.04)	0.10^**	(0.04)	0.10^**	(0.04)
Rural & Two Generation	−0.06	(0.03)	−0.02	(0.05)	−0.14^**	(0.05)	−0.14^**	(0.05)
Rural & Three Generation	0.01	(0.04)	0.03	(0.06)	−0.13	(0.07)	−0.13	(0.07)
Rural & Skip Generation	0.11^**	(0.04)	0.20^**	(0.06)	0.05	(0.07)	0.05	(0.07)
Rural & Living Alone	0.05	(0.03)	0.09	(0.05)	0.07	(0.06)	0.07	(0.06)
Rural & All Other Arrangements	0.03	(0.05)	0.10	(0.07)	−0.07	(0.07)	−0.07	(0.07)
Observations	12,674		12,655		12,629		12,836
Panel B. Among 65+
Urban & Living with Spouse Only	Reference	Group	Reference	Group	Reference	Group	Reference	Group
Urban & Two Generation	−0.14^**	(0.05)	−0.25^***	(0.07)	−0.14	(0.08)	−0.02	(0.07)
Urban & Three Generation	−0.06	(0.05)	−0.18^*	(0.08)	0.01	(0.10)	−0.03	(0.09)
Urban & Skip Generation	−0.09	(0.08)	−0.25^*	(0.13)	−0.15	(0.13)	0.09	(0.13)
Urban & Living Alone	−0.07^*	(0.04)	−0.17^**	(0.06)	−0.03	(0.07)	−0.04	(0.06)
Urban & All Other Arrangements	−0.12	(0.12)	−0.32^*	(0.13)	−0.05	(0.18)	0.02	(0.16)
Rural & Living with Spouse Only	0.05	(0.04)	0.13^*	(0.06)	0.12	(0.06)	−0.08	(0.05)
Rural & Two Generation	−0.18^***	(0.05)	−0.22^**	(0.08)	−0.15	(0.09)	−0.20^**	(0.07)
Rural & Three Generation	−0.11	(0.06)	−0.17	(0.10)	−0.09	(0.11)	−0.15	(0.09)
Rural & Skip Generation	0.02	(0.07)	0.18	(0.12)	0.02	(0.15)	−0.11	(0.09)
Rural & Living Alone	−0.00	(0.06)	0.03	(0.09)	0.00	(0.09)	−0.12	(0.07)
Rural & All Other Arrangements	−0.04	(0.06)	0.06	(0.10)	−0.17	(0.11)	−0.21^**	(0.08)
Observations	5,100		5,093		5,082		5,154

Note. Standard errors in parentheses. Models controlled for all other covariates as shown in Table 2.

*p < .05. **p < .01. ***p < .001.

Results in Table 5 suggest that, among those ages 50+, the adverse association between two-generation households and SWB did not differ by hukou status. However, older adults with a rural hukou living with a spouse only reported significantly higher SWB in all domains than their counterparts with an urban hukou. Rural hukou status was also associated with significantly higher global SWB and life satisfaction for those in skipped-generation households than for those with an urban hukou living with a spouse only. In contrast, results for those aged 65+, shown in Panel B of Table 5, suggest that a rural hukou and living in two-generation households were associated with significantly lower SWB, including global SWB, life satisfaction, and feelings of control.

Discussion

Our results add to our understanding of how living arrangements relate to different domains of SWB and show that context matters. Across the various analyses, our findings suggest that living with one’s spouse promotes SWB. Both living alone and in “other” arrangements were associated with lower life satisfaction and happiness than living with a spouse only. Consistent with theories and prior empirical evidence, resources play a vital moderating role in the complex relation between living arrangements and SWB, a finding that could help inform prevention and intervention efforts in China.

Chinese cultural norms suggest that living with adult children is likely to promote SWB. Our results, however, indicate that living in two-generation households might not be associated with higher SWB for older adults compared to those living with a spouse only. Empirical studies have found that living in a preferred arrangement is more important for older adults’ SWB than following societal norms (Chen, 2019). A growing percentage of contemporary Chinese older adults may prefer to live with their spouses only. Previous studies have also suggested that living with adult children may lead to reduced privacy and increased intergenerational conflict, both of which can put stress on relationships (Hermalin & Yang, 2004). Moreover, the adverse association between two-generation households and SWB was most pronounced among adults in both 50+ and 65+ samples with income in the bottom third of the distribution. Although beyond the scope of this paper, we speculate that the challenges of living in or near poverty may exacerbate the strain on relationships when living in a multigenerational household. Older adults in these low-income households may need to contribute more physically, emotionally, and financially to the maintenance of the household than their higher income counterparts.

Chinese cultural norms suggest that older adults’ SWB benefit from multigenerational living arrangements, particularly living with both adult children and grandchildren. Previous studies have confirmed this positive association (Burnette et al., 2013; Chen, 2019; Zhang, 2015). Our results on three-generation households versus living with only a spouse are mixed. We found that global SWB, life satisfaction, and happiness were lower among those living in three-generation households than among those living with their spouse only. However, this negative association applied only to the life satisfaction and happiness of those in the bottom third of the income distribution and to the life satisfaction of those aged 65+ with an urban hukou. These findings may again indicate that the costs of a multigenerational household under resource-constrained circumstances might outweigh the benefits.

Are two- and three-generation living arrangements truly not conducive to older adults’ SWB? One consistent result from this analysis is a strong positive association between living with a daughter (vs. living without) and SWB, including life satisfaction and happiness. In addition, living with a daughter-in-law (vs. without) also showed a strong positive association with feelings of control, though to a lesser degree. These findings seem to refute the adverse association between two-generation living arrangements and SWB. Chen and Short (2008) found that living with daughters indeed promotes emotional health among older adults, more so than living with sons, a result at odds with the cultural norm designating sons as the primary caregivers. Possibly, the kind of care provided by daughters is qualitatively different from that of sons and that the emotional bond between parents and daughters helps promote SWB (Chen & Short, 2008).

In addition, our results regarding daughters-in-law, albeit relatively weaker than those related to daughters, may indicate that patrilineal culture in China requiring daughters-in-law to care for her husband’s parents may still be relevant in contemporary China. The positive associations between SWB and living with daughters and daughters-in-law, together with a non-significance associated with sons/sons-in-law, highlight the importance of gender-patterned caregiving and household dynamics in assessing older adults’ SWB (Cong & Silverstein, 2008).

We also found that those living in skipped-generation households, on average, did not have significantly different SWB compared with those living with a spouse only, except that those aged 50+ had significantly stronger feelings of control. Importantly, older adults in the middle or top third of the income distribution living in skipped-generation households reported having significantly higher SWB in multiple domains, but having significantly lower life satisfaction when they resided in urban areas. In line with research underscoring the importance of access to resources among older adults, we expected older adults residing in urban areas to generally have better SWB than their rural counterparts. Instead, rural adults aged 50+ living in skipped-generation households had significantly higher SWB (0.11 vs. −0.04, p < .01) and life satisfaction (0.20 vs. −0.19, p < .001). Also, rural adults aged 50+ living only with a spouse had significantly higher SWB in all domains than their urban counterparts (but the association was weaker for those aged 65+). This rural–urban divide in global SWB and life satisfaction might be explained, in part, by differences in social interactions and physical activity between the two groups. Yi et al. (2001) found that older adults living in rural China were healthier and more active, and reported generally better quality of life than their urban counterparts. It is likely that these older adults residing in rural areas living with only grandchildren were a select group of physically healthy adults. Yet, our analysis controlled for physical health and other important selection factors, such as education and income. We speculate, although beyond the scope of this paper, that older adults in rural skipped-generation households may have been economically supported by their adult children, thereby bolstering the life satisfaction of the grandparents (Li & Yang, 2017). It is also possible that the emotional bonds with grandchildren, together with a sense of obligation to raise the next generation, may provide a feeling of accomplishment that directly translates into positive perceptions of life. This might be particularly true for those living in rural areas, a reflection of the urban-rural differences in family norms and expectations (Chen, 2019). Massive internal migration of young adult children has reframed grandparenting to include more childrearing responsibilities, but modernization has left rural areas more rooted in traditions and cultures than urban areas (Chen, 2019). In contrast, for urban older adults in skipped-generation households, the discrepancy between their aspirations and reality may be more acute, increasing the stress of such an arrangement. We note, however, that adults aged 65+ with rural hukou status living in two-generation households had significantly lower feelings of control than their urban counterparts (−0.02 vs. −0.20, p < .05).

The rural-urban differences in our results underscore the importance of hukou status in understanding the association between living arrangements and SWB. Living arrangements represent an interplay between cultures, beliefs, and resources, which shapes both one’s preferred and de facto living arrangements. Hukou status, established originally to be an institutional context to control one’s residence and movement, now defines socioeconomic context. It is also a cultural factor (e.g., through family norms and expectations) that not only shapes but also moderates how ones’ living arrangements may be associated with SWB.

Limitations

As with any observational study, we note several limitations associated with this analysis. First, we considered a broad set of SWB domains corresponding to affection and cognition. However, our data may not allow us to fully capture each dimension of SWB. For example, our results regarding feelings of control depended on two items; a more comprehensive scale like the Pearlin Mastery Scale (Pearlin & Schooler, 1978) might be better suited to examining this aspect of SWB. Second, although we categorized living arrangements in as much detail as possible, due to sample sizes, we were unable to distinguish older adults living with or without spouses in multigenerational households.

Third, small sample sizes also prevented us from further dividing the sample by those aged 75 and over. Recent studies have found that whereas people aged 50–64, on average, had better-than-expected SWB, those aged 75+ may suffer from low SWB, primarily due to loneliness brought on by the death of spouses and friends. Loneliness carries significant consequences for health, well-being, and cognition (Hawkley et al., 2019; Suanet & van Tilburg, 2019).

Fourth, some measures are far from perfect, such as those for smoking and drinking. More detail such as the frequency of drinking could help explain variations in SWB. Similarly, the SAGE data only collected characteristics of adult children (e.g., age, education, marital status, health status) who were living with older adults at the time of the survey. We were thus unable to use these data due to our interest in comparing SWB of those living alone or in skipped-generation households with those in two- or three-generation households.

Finally, our results only show associations, not causality. Statistical approaches such as using an instrumental variable (IV) could produce estimates one step closer to causality. An IV approach requires identification of extraneous variables that are directly related to living arrangements but not to SWB; the only path linking extraneous variables to SWB would be indirectly through living arrangements. Information such as the number of adult children or the age of the oldest adult child would be ideal IVs (Do & Malhotra, 2012). However, again, such information was only collected for older adults who lived with their adult children, limiting our ability to conduct such an analysis for all living arrangements. Despite this, our findings are largely consistent with theories and existing empirical evidence showing the importance of living arrangements. These limitations demonstrate the need for more refined analyses using large longitudinal datasets with better data on living arrangements to deepen our understanding of how living arrangements, a crucial socio-cultural factor, may matter to SWB.

Conclusion

In sum, a substantial share of adults aged 50 or over in our sample lived with a spouse only, the group with the best overall SWB. Our mixed results on multigenerational living arrangements underscore the importance of context, consistent with tenets of social stratification theory. Our results, together, stress the importance of resources and preferences. When physically and financially feasible, older adults in contemporary China might prefer to live with their spouses instead of following cultural norms, translating to better SWB. We also found that resources act as a significant protective factor across living arrangements. Our results on the interplay between income and living arrangements may suggest an important direction for policy development to support multigenerational families in need. The gap in SWB associated with resource access across income groups is particularly concerning given China’s increasing socioeconomic inequality (NBSC, 2015). Lower levels of SWB among older adults have been linked to poorer functional health, less social engagement, and higher rates of depression and mortality (Huxhold et al., 2013). Limited socioeconomic resources have been shown to be a strong predictor of poor physical and psychological health (George, 2010). From a public health perspective, it will be critical for both clinical practitioners and policymakers in China to consider the population-level impacts of older adults experiencing low levels of SWB associated with living with adult children. Improving socioeconomic parity by ensuring resources for older adults could be an important preventative investment in the health and well-being of Chinese society.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Wen-Jui Han

Ying Li

Author Biographies

Wen-Jui Han, professor of Social Work at New York University and co-editor-in-chief of the International Journal of Social Welfare. With substantive knowledge and skills from multidisciplinary training in sociology, developmental psychology, public policy, and social work. His area of expertise is on social welfare policy with an emphasis on families and paying particular attention to vulnerable and marginalized populations.

Ying Li is an associate professor at the Renmin University of China's School of Labor and Human Resources. Her research interests include elder welfare and disability policy.

Cliff Whetung is a PhD student at the New York University's Silver School of Social Work. His research interests include Indigenous cognitive health, intergenerational services, and discrimination effects on the health of older adults.

References

Burnette

Sun

(2013). A comparative review of grandparent care of children in the U.S. and China. Ageing International, 38, 43–57. https://doi.org/10.1007/s12126-012-9174-z

Chen

Short

S. E.

(2008). Household context and subjective well-being among the oldest old in China. Journal of Family Issues, 29(10), 1379–1403.

Chen

Han

W-J.

(2016). Front-runner of community-based elder care in China. Journal of Aging & Social Policy, 28(4), 292–307.

Chen

(2019). Living arrangement preferences and realities for elderly Chinese: Implications for subjective well-being. Ageing & Society, 39, 1557–1581.

Chen

Silverstein

(2000). Intergenerational social support and the psychological well-being of older parents in China. Research on Aging, 22(1), 43–65.

Chow

. (2009). Filial piety in Asian Chinese communities. In Sung

Kim

B. J.

(Eds.), Respect for the elderly: Implications for human service providers (pp. 319–323). University Press of America.

Cong

Silverstein

(2008). Intergenerational support and depression among elders in rural China: Do daughters-in-law matter? Journal of Marriage and Family, 70(3), 599–612.

Diener

(1994). Assessing subjective well-being: Progress and opportunities. Social Indicators Research, 31(2), 103–157.

Y. K.

Malhotra

(2012). The effect of coresidence with an adult child on depressive symptoms among older widowed women in South Korea: An instrumental variables estimation. The Journal of Gerontology, Series B: Psychological Sciences and Social Sciences, 67(3), 384–391.

10.

Dolan

Metcalfe

(2012). Measuring subjective well-being: Recommendations on measures for use by national governments. Journal of Social Policy, 41(2), 409–427.

11.

Feng

Liu

Guan

Mor

(2012). China’s rapidly aging population creates policy challenges in shaping a viable long-term care system. Health Affairs, 31(12), 2764–2773.

12.

Ferring

Boll

. (2010). Subjective well-being in older adults: Current state and gaps of research. In Bovenberg

Soest

A. V.

Van Soes

A. H. O.

(Eds.), Ageing, health and pensions in Europe (pp. 173–212). Palgrave Macmillan.

13.

(2014, May 30). China National Committee on Aging will conduct “Fourth Urban-Rural Aging Population Sample Survey in China” in 2015. http://www.cncaprc.gov.cn/contents/2/3866.html

14.

George

L. K.

(2010). Still happy after all these years: Research frontiers on subjective well-being in later life. Journal of Gerontology: Social Sciences, 65B(3), 331–339.

15.

Greene

W. H.

(2012). Econometric analysis (7th ed.). Pearson Education.

16.

Guo

. (2016). Understanding fertility trends in China. In Guilmoto

C. Z.

Jones

G. W.

(Eds.), Contemporary demographic transformations in China, India and Indonesia (pp. 97–111). Springer International.

17.

Hawkley

L. C.

Wroblewski

Kaiser

Luhmann

Schumm

L. P.

(2019). Are U.S. older adults getting lonelier? Age, period, and cohort differences. Psychology and Aging, 34(8), 1144–1157.

18.

Hermalin

A. I.

Yang

(2004). Levels of support from children in Taiwan: Expectations versus reality, 1965–99. Population and Development Review, 30(3), 417–448.

19.

Huxhold

Fiori

K. L.

Windsor

T. D.

(2013). The dynamic interplay of social network characteristics, subjective well-being, and health: The costs and benefits of socio-emotional selectivity. Psychology and Aging, 28(1), 3–16. https://doi.org/10.1037/a0030170

20.

S.S.

Yang

(2017). Ge dai fu yu zi nv yang lao xing wei xuan ze: shi yu jie shi [Skip generational care and support from adult children: facts and explanations]. Statistics & Information Forum, 32(7), 102–107.

21.

Lin

E. H.

Katon

Von Korff

Tang

Williams

J. W.

Jr Kroenke

Hunkeler

Harpole

Hegel

Arean

Hoffing

Della Penna

Langston

Unützer

IMPACT Investigators. (2003). Effect of improving depression care on pain and functional outcomes among older adults with arthritis: a randomized controlled trial. Journal of the American Medical Association, 290(18), 2428–2429.

22.

Little

R. J. A.

(1988). A test of missing completely at random for multivariate data with missing values. Journal of the American Statistical Association, 83, 1198–1202.

23.

Ministry of Civil Affairs, People’s Republic of China. (2017, August 3). Statistical communique of the People’s Republic of China on the 2016 social services development. http://www.mca.gov.cn/article/sj/tjgb/201708/20170800005382.shtml

24.

National Bureau of Statistics of China. (2015). China statistics yearbook 2015. China Statistics Press. http://www.stats.gov.cn/tjsj/ndsj/2015/indexeh.htm

25.

A. C. Y.

Phillips

D. R.

Lee

W. K-M.

(2002). Persistence and challenges to filial piety and informal support of older persons in a modern Chinese society: A case study in Tuen Mun, Hong Kong. Journal of Aging Studies, 16(2), 135–153.

26.

Nieminen

Lehtiniemi

Vähäkangas

Huusko

Rautio

(2013). Standardised regression coefficient as an effect size index in summarising findings in epidemiological studies. Epidemiology, Biostatistics and Public Health, 10, e8854.

27.

Ohrnberger

Fichera

Sutton

(2017). The relationship between physical and mental health: A mediation analysis. Social Science & Medicine, 195, 42–49.

28.

Pearlin

L. I.

Schooler

(1978). The structure of coping. Journal of Health and Social Behavior, 19, 2–21.

29.

Pinquart

Sorensen

(2000). Influences of socioeconomic status, social network, and competence on subjective well-being in later life: A meta-analysis. Psychology and Aging, 15, 187–224.

30.

Ryff

C. D.

Singer

(1998). The contours of positive human health. Psychological Inquiry, 9(1), 1–28.

31.

Silverstein

Cong

(2006). Intergenerational transfers and living arrangements of older people in rural China: Consequences for psychological well-being. Journals of Gerontology: Psychological Sciences and Social Sciences, 61B(5), S256–S266.

32.

Suanet

van Tilburg

T. G.

(2019). Loneliness declines across birth cohorts: The impact of mastery and self-efficacy. Psychology and Aging, 34(8), 1134–1143.

33.

Tang

T. Y.

(2016). Wo guo lao nian ren ju zhu an pai zhuang kuang fen xi——ji yu 2012 nian zhong guo jia ting zhui zong diao cha shu ju [An analysis of the living arrangements of the elderly in China: The Longitudinal Survey of Chinese Families in 2012]. South China Population, 31(4), 50–58.

34.

United Nations. (2019). World population prospects, 2019, volume 1, comprehensive tables (ST/ESA/SER.A/426). Department of Economic and Social Affairs, Population Division. https://population.un.org/wpp/Publications/Files/WPP2019_Volume-I_Comprehensive-Tables.pdf

35.

von Hippel

P. T.

(2007). Regression with missing Ys: An improved strategy for analyzing multiply imputed data. Sociological Methodology, 37, 83–117.

36.

Wang

Y. S.

(2014). Zhong guo cheng xiang lao nian ren ju zhu de jia ting lei xing yan jiu——ji yu di liu ci ren kou pu cha shu ju de fen xi [A study on the type of household structure among the elderly in China: The analysis of the 6th census]. Chinese Journal of Population Science, 1, 22–34.

37.

White

I. R.

Royston

Wood

A. M.

(2011). Multiple imputation using chained equations: Issues and guidance for practice. Statistics in Medicine, 30, 377–399.

38.

Zheng

(2018). Household registration, urban status attainment, and social stratification in China. Research in Social Stratification and Mobility, 53, 40–49. https://doi.org/10.1016/j.rssm.2017.11.002

39.

Z. Q.

Sun

Y. H.

Zhang

X. J.

Tao

F. B.

Cui

G. H.

(2010). Correlation between loneliness and social relationship among empty-nest elderly in Anhui rural area, China. Aging & Mental Health, 14(1), 108–112.

40.

Burnette

Cheng

(2017). Living arrangements and subjective well-being among older adults in China. Innovation in Aging, 1(1), 377.

41.

Vaupel

J. W.

Zhenyu

Chunyuan

Yuzhi

(2001). The healthy longevity survey and the active life expectancy of the oldest old in China. Population: An English Selection, 13(1), 95–116.

42.

Zhang

(2015). Living arrangements and subjective well-being among the Chinese elderly. Open Journal of Social Sciences, 3, 150–161.

43.

Zhou

Wang

Liu

Zhang

Zheng

(2015). Socio-economic factors related with the subjective well-being of the rural elderly people living independently in China. International Journal for Equity in Health, 14(1), 5.

44.

Zhu

W. X.

(2003). The one-child family policy. Archives of Disease in Childhood, 88, 463–464. https://doi.org/10.1136/adc.88.6.463