An Ecological Framework for Active Aging in China

Abstract

Objectives: Against the background of emerging research interest in integrating Active Aging in long-term care policies, the study investigates the relationships between community environment and engagement in social activities (paid work, domestic care, participation in community and leisure activities) by older Chinese adults. Six indicators are derived from the economic, institutional, and sociodemographic environments of the communities under analysis. Method: The first wave of the China Health and Retirement Longitudinal Study (CHARLS) is used for empirical examinations. The sample includes respondents aged 50 years and above (n = 6,290) from 307 communities. Random-intercept multilevel logistic models are employed. Results: Economic environment is positively linked to engagement in paid work and grandparental childcare, community activities, and leisure activities. However, the effects decreased after including the indicators for the institutional and sociodemographic environments. We found that older adults are more likely to engage in community activities and leisure activities in communities and with a higher percentage of migrant population. Discussion: To integrate Active Aging in the current long-term care policies in China, the study points out that residential community environment is crucial for older Chinese to enhance and sustain their involvement in family and communities before older adults rely on long-term care facilities. In particular, the relevance of institutional and sociodemographic environment provides policy makers to rethink about how to provide community-based long-term care.

Keywords

active aging ecological approach multilevel analysis China communities long-term care

Introduction

Long-term care has been the main focus of policy considerations for providing systematic resources and support to meet the care needs of older adults in late life. Given the fact that the advancement of medicine not only prolongs life expectancy but also delays health decline, the focus of care has been shifting gradually in some countries. The interest in incorporating long-term care with the social behavior dimension, called Active Aging, to promote healthy aging is receiving increasing policy attention, such as in the European Union (Rodrigues, Hofmarcher, Leichsenring, & Winkelmann, 2013). It addresses: “the process of optimizing opportunities for health, participation and security in order to enhance quality of life as people age” (World Health Organization, 2002). This process involves socially active engagement in family, social and professional life, including engagement in paid work, domestic care, community activities, and leisure activities (Avramov & Maskova, 2003; World Health Organization, 2002) because activity engagement is associated with better well-being and with delaying declining health in later life (Gessa & Grundy, 2014; Herzog & Morgan, 1992; Riley & Riley, 1994). The present study focuses on China, one of developing societies that face rapid demographic aging. Shrinking household structures due to the one-child policy and labor migration have led to increasing concerns about older adults’ care and support (N. J. Zhang, Guo, & Zheng, 2012). The government has developed long-term care policies for the foreseeable rapid aging population by expanding medical institutions, upgrading health care resources and building community-based care facilities (Feng et al., 2011; State Council, 2001; Wu, Carter, Goins, & Cheng, 2005). Nevertheless, the current focus of the long-term care remains to concentrate on meeting the care need and provision of medical resources to old frail adults. Some scholars have addressed the policy framework of promoting activity engagement in employment, volunteer activities and care provision (Du & Yang, 2010; Lum, 2013; Morrow-Howell & Wang, 2013). Up to now, research has provided findings on individual determinants of different activity engagement among older Chinese (cf. H. Li, Chi, & Xu, 2010; H. Liu & Lou, 2016). As community support and resources for older population have been emphasized in its recent community-building project (Shen, 2014; Yeatts, Cready, Pei, Shen, & Luo, 2014), few research investigate the extent of community environment influence older adults’ activity engagement.

The present study uses an ecological approach to examine impacts of community environments on activity engagement among older Chinese. Empirical studies based on Western developed economics have made use of the ecological approach to investigate how built-in and social components of communities provide opportunities or constrain motivation and engagement (Dury et al., 2014; Paillard-Borg, Wang, Winblad, & Fratiglioni, 2009). As Morrow-Howell and Wang (2013) emphasize that the including sociocultural contexts is crucial to advance conceptual frameworks and social policies for aging populations cross-culturally, our study contributes to the current literature on Active Aging in China.

An Ecological Approach to Activity Engagement in Communities in China

Earlier literature has developed ecological models to understand the extent of community environment impacts on older adults’ well-being as older adults spend more time in their communities than the middle-aged do (Day, 2008; Lawton, 1982). An ecological model emphasizes that when the fit between older adults’ competence (individual resources) and their communities is balanced, older adults are more likely to participate in activities (Lawton, 1982; Rowles & Bernard, 2013). It considers communities not merely as space for older adults to reside in, but as space that affects their social relationships and psychological well-being (Annear et al., 2014; Lawton, 1982; Wahl & Weisman, 2003). In communities where there is an imbalance between older adults’ competence and their community characteristics, the older adults are required to use their own resources to fill the gap to join activities, increasing their vulnerability. Communities can be deconstructed into various community environments, which enable older residents to accomplish tasks or engage in desired behavior, even after individual factors are taken into consideration (Glass & Balfour, 2003; Lawton, 1982). For example, if a community does not have access to public transportation, travel will be a constraint for older adults’ with disabilities and thus their motivations to go out will be reduced. On the contrary, communities that provide accessible resources and infrastructure will facilitate social activities (Dury et al., 2014; Richard, Gauvin, Gosselin, & Laforest, 2009). The multidimensional community environment includes not only physical features of the communities, but also the compositional features that influence older people’s perceptions of their communities, which explain why older adults’ behavior and mental well-being would be affected by the community in which they reside.

Communities in China are collective units for mobilizing locals and advocating government policies, unlike the grassroots nature of communities in western countries (Bray, 2006). During the economic reforms of the 1980s, the government initiated the community-building project (Xu, 2007), in which the “urban street committee” (juweihui in Chinese) became independent from government administration (Guan & Chow, 2003), while the “rural village committee” (cunweihui in Chinese) took charge of village matters (X. Zhang, Fan, Zhang, & Huang, 2004). Committees became units to delegate welfare responsibilities of the state to other potential social sectors locally (Yan & Gao, 2007) and are made responsible for channeling welfare resources among its members (Xu & Chow, 2006).

Limited research conceptualizes dimensions of community environments to study activity engagement among older adults in China. Some studies, which applies the ecological approach to health outcomes of older adults point out that regional economic development and institutional support represent important community resources for older adults (L. W. Li, Liu, Xu, & Zhang, 2016; Shen, 2014; Yeatts et al., 2014; Yeatts et al., 2013). Moreover, studies based on the western contexts show that the residential composition is relevant to the activity engagement as it can affect the atmosphere and preferences of engagement (Okun & Michel, 2006; Wilson, 2000). Thus, we review the literature on the economic, institutional and sociodemographic environments of communities

Economic, Institutional, and Sociodemographic Environments of Communities

In western contexts, a lower income scale at the community level for older adults is found to be negatively associated with volunteering (Dury et al., 2014; Hussein & Manthorpe, 2014). A community that faces income deprivation would more likely be associated with low trust among neighbors due to its likelihood of having a higher crime rate and disordered environment (Dury et al., 2014; Wilson, 2000). Although economic prospects and governmental investment have provided urban areas with better facilities and services in both the public and private sectors (Zimmer, Kaneda, & Spess, 2007; Zimmer, Wen, & Kaneda, 2010), rural areas are less developed and lack basic infrastructure (Yeatts et al., 2013). These relative deprivations lead to the situations that older adults in rural areas may not have resources and services from public and private sectors (L. W. Li et al., 2016; Park, 2008). Hence, urban–rural differences may indirectly influence the opportunities for older adults to engage in the activities under the categories of Active Aging. Economic development affects the job opportunities of paid work, types of economic activities as well as the community resources for older adults (M. Li et al., 2013; Pang, De Brauw, & Rozelle, 2004). H. Liu and Lou (2016) point out that the concept of formal volunteering is less recognized in rural China, which implies that older adults in rural areas may not plan to join formal volunteering or perceive volunteering as a part of potential later-life activities. Hence, we expect that the following:

Hypothesis (H1): Older adults in urban areas are more likely to engage in activities, whereas older adults in rural communities are less likely to do so.

The institutional environment of a community refers to the purposely established environment by the government as the role of government is a crucial “third party” to set up community capacities and built-in environment (Haski-Leventhal, Meijs, & Hustinx, 2010; Henkin & Zapf, 2006-2007). The community governance of China reflects the features of the top-town recognition and institutional environment of communities in building community physical capacities, channeling welfare resources and providing social services (Bray, 2006; Saich, 2000). Thus, we use the term “institutional environment” to signify the governmental role. For economic activities, institutional environment can be overlapped with the economic development as it reflects the local investment. For nonpaid activities, physical space, organization, venues and settings can indirectly affect the social cohesion in communities because these elements are important community capacities that generate the occurrence of social connections and open up opportunities for people to participate in activities, especially volunteering (Henkin & Zapf, 2006-2007; Rotolo & Wilson, 2012). For older adults, sufficient institutional environment will compensate for their decreasing or lost competencies (e.g., physical disabilities), support them to sustain their social inclusion, and enable them to participate in activities in the communities (Wahl & Weisman, 2003). Community infrastructure and facilities are regarded as two indicators in the analysis because both indicators increase older adults’ physical participation and facilitate a sense of community, allowing residents to feel socially included (Henkin & Zapf, 2006-2007; Wahl & Weisman, 2003). Even though the rural–urban divide is relevant to the variations in the institutional environment across communities, we argue that the institutional environment needs to be separated from the economic environment because it enables detailed examination concerning the institutional establishments, not just the economic disparities. Hence, based on these explanations, we hypothesize that

Hypothesis (H2): Older adults in communities with better community infrastructure are more likely to engage in activities than those in communities with poorer community infrastructure.

Hypothesis (H3): Older adults in a community with more public facilities are more likely to engage in activities than those who live in communities with fewer public facilities.

The sociodemographic environment refers to the socioeconomic- and demographic-related characteristics of residents in a community. These characteristics may influence the social process of activity engagement, which are addressed by research on volunteering and community activities: Education is associated with individuals’ awareness of collective problems, activeness and skills they possess to engage in volunteering (Wilson, 2000). That is, when a community with higher number of residents with better education, residents would be more like to sustain or improve neighborhood quality by dealing with issues through public discussions and volunteering activities (Musick & Wilson, 2008; Wilson, 2000). When less educated individuals are exposed to an environment where educated residents are involved in volunteering, they may develop a sense of belonging to the community and be more likely to participate in volunteering (Dury et al., 2014; Okun & Michel, 2006). Using the similar logic, older adults living in a community with a high percentage of educated residents will be aware of and be used to the idea of working in later life to sustain social connections. The percentage of residents with a college degree is chosen as an indicator:

Hypothesis (H4): Older adults in a community with higher percentage of college-educated residents are more likely to participate in activities.

The second indicator is the percentage of older people in a community. This indicator reveals the degree of population aging at the community level (Dury et al., 2014). We argue that living in a community with a higher degree of population aging may indirectly affect older residents’ perceptions of themselves as being physically abandoned in communities, as the massive “labor migration” in China—young people go to cities for better opportunities, leaving their aged parents in their hometowns—has been documented in research (cf. Zimmer & Kwong, 2003). A community with a high percentage of older population implies that it is a community that young people have left for work. Hence, having fewer young people around in the community may create a psychological barrier for older adults that restricts them from becoming more active in taking up social activities. We anticipate that

Hypothesis (H5): older adults are less likely to engage in activities in communities in which a higher percentage of the population is older.

The last indicator is related to the cohesion of a community In a residential-stable community, residents know each other well and are more willing to put in effort to improve the community environment (e.g., Swaroop & Morenoff, 2006; Wilson, 2000). For older people, feeling secure and stable in their communities is relevant to their engagement in volunteering (Dury et al., 2014; Okun & Michel, 2006). In China, labor migration from rural to urban areas indicates a push–pull force of young workers that changes residential stability and composition in villages and cities (A. Chen & Coulson, 2002). Living in a community with higher number of migrants may decrease their motivation to engage in activities. Hence, the last hypothesis is derived:

Hypothesis (H6): Older adults in communities with a higher percentage of migrant population are less likely to participate in activities.

Method and Data

Data

The data used for this study comes from the first wave of the 2011 China Health and Retirement Longitudinal Study (CHARLS), which is designed after the Health and Retirement Study (HRS) of the older American population. Through multistage sampling, it was implemented in 2011 to cover 28 Chinese provinces and surveyed a sample of 17,587 noninstitutionalized Chinese (above 45 years of age) in about 10,250 households in 150 counties/districts as a nationally representative survey (Zhao et al., 2013). The response rate was approximately 80.5%. CHARLS employed a separate community survey that collected information from the person in charge of the selected community such as village head or director of the street committee about basic demographics, community infrastructure, facilities, history of policies, and so on (Zhao et al., 2013). 450 village/resident committees in which the CHARLS collected individual characteristics were covered in the community survey and hence we are able to link information of individual characteristics to the community characteristics of their residence for the analyses.

The analytic sample was restricted to respondents aged 50 years and above with at least one grandchild. The eligibility age for retirement in China is 60 for men and 55 for women (T. Liu & Sun, 2016) and hence the sample can capture those who are close to the retirement age. The CHALRS is a household-based survey, which surveyed both main respondents and their spouses (if age eligible). To reduce the impacts of methodological interdependency when including the spouses’ information, we only include the main respondents. Furthermore, limiting to respondents with at least one grandchild enables us to examine engagement in domestic childcare. After dropping cases with missing values, the analytic sample totaled 6,290 respondents from 307 communities (123 urban communities and 184 rural communities).

Variables

The dependent variables consist of three kinds of activity engagement addressed in Active Aging literature. The first activity is paid work. If a respondent reported engaging in paid work (nonagricultural) the week before, it is coded as one. The second activity is domestic care and we focus on grandparental childcare when a respondent reported taking care of a grandchild below age of 16 years within the last year, it is coded as one.¹ The third is engagement in community life and leisure activities. It is coded as one if a respondent reported having engaged in the following activities in the previous month: helping family, friends, or neighbors, voluntary or charitable work, caring for a sick or disabled adult, playing mahjong, chess, cards, going to community clubs, taking part in a sport, social or other kind of club, or attending an education or training course. We include agricultural activities for additional analyses, considering that China’s developing economy, which agricultural activities are still important economic activities.

Indicators of the three community environments are constructed: We distinguish urban/rural communities from information provided by the National Bureau of Statistics. For the institutional environment, two indicators are constructed. First, the infrastructure sufficiency index is constructed based on the principal component analyses: we consider the following conditions: a community has paved roads, refuse removal, a sewerage system, toilets with water or without water, drinking water, and natural gas or liquefied petroleum gas as fuel (in terms of the percentage of households using these two types of fuel). This is a continuous variable and the Cronbach’s alpha is .69. The second indicator is ratio of public facilities, which is constructed by first totaling the number of public facilities including kindergartens, primary schools, junior middle schools, senior high schools, post offices, libraries, police stations, banks, theaters, nursing homes, convenience stores, farmers’ markets, and supermarkets. As larger communities may have more public facilities, the total was divided by the number of residents and then multiplied by 100. We created two categories (0-0.40 and >0.41) to have almost equally 50% of communities in one of each category as the distribution is right skewed. For the sociodemographic dimension, the first indicator is the percentage of population aged 65 and above (categorical variable = 0%-11%, 12%-25%, or >25%). The CHARLS community questionnaire has the number of the total population in the village or community for more than half a year at the end of the last calendar year (2010) and the number of older population aged 65 years and above. We divided the number of older population aged 65 years and above by the total population of residents and categorized the percentages into three groups (0%-11%, 12%-25%, and >25%). The second indicator is the percentage of college-educated residents. It is a categorical variable (0%-1%, 1%-5%, or above 5%) as we merge the categories from the question in the CHARLS community questionnaire: “What percentage of adult population have completed only up to college? (1) No, (2) 0%-1 %, (3) 1%-5%, (4) 5%-10%, (5)10%-15% and (6) >15%.” The last indicator is percentage of migrant population. The CHARLS community questionnaire has the number of total migrant population (who lived less than half a year in the surveyed village or community) at the calendar year (2010). We divided the number of migrant population by the total number of residents (also in the community questionnaire) to calculate the percentage of migrant population. Then, we categorized the number into three groups: 0, 0%-5%, and 5% above to compare the differences of communities in different percentages of migrant population.

The study controls for individual characteristics. Age (four categories: 50-59 years, 60-69 years, 70-79 years, and above 80 years), gender (binary variable; 1: female), having a partner (binary variable: 1 = yes), activities of daily living (ADL) disabilities, such as eating and bathing (binary variable: 1: having at least one disability). Education is measured based on four categories of education level (illiterate, capable of writing and/or reading, elementary school/home school, and middle school and above). For financial resources, we construct household expenditure per capita because income can vary through the years, especially for rural households (e.g., Lei, Sun, Strauss, Zhang, & Zhao, 2014). Hence, household expenditure per capita is summed up by expenditures for food, dining out, alcohol, cigarettes, communication fees, utilities, transportation, taxes, and so on. The sum is divided by the number of coresident family members and then is calculated by logarithm. Living with at least one child is coded as a binary variable. Receiving a pension or not is coded if respondents reported receiving supplemental pension insurance from employer, a commercial pension, rural pension, residents’ pension, urban residents’ pension, or pension subsidy to the oldest old (binary variable: 1 as receiving pension). To capture whether older adults’ have migration experience, a binary variable, Hukou, changed is coded when the respondents changed (a) their Hukou types (agricultural and nonagricultural Hukou) from their first Huko type, (b) change Hukou places (different from their birth places). We deal with missing values in individual characteristics and community characteristics by using listwise deletions. The scores of variance inflation factor (VIF) for all these variables are between 1 and 2, so the scores indicate the least concern about collinearity.

Analytic Approach

Given that the nature of the data includes the concept that individuals are embedded within communities, along with the use of binary coding for activities, two-level random-intercept multilevel models for binary data are employed.²

Three kinds of activities constructed are analyzed separately. After confirming that there are significant community variations, we interpret the results of two-way models. Intraclass correlation coefficients (ICC, denoted as ρ) are provided at the end of the tables to examine whether adding variables in nested models decreases the between-community variances. Four consecutive models are performed to examine effects on activities: Model 1 includes control variables of individual characteristics. Variables related to community environments are specified in Models 2, 3, and 4 with indicators of the economic, institutional and sociodemographic environments being added accordingly. The purpose of this order is to explore whether effects of indicators from institutional and sociodemographic environments are robust as the economic environment is closely associated with the development of community capacities and sociodemographic features of a community. Additional analyses are performed with individual weights in the Appendix (the CHARLS does not provide community weights).

Results

Table 1 shows that the analytic sample consisted of 8.4% of older adults who reported to have paid work, 22.0% of them who provided childcare to their grandchildren, and 22.6% in in community and leisure activities. Considering the participation rate of agricultural activities is 49.2% (not shown), we decided to include the analyses on agricultural activities in the additional analyses. The age distribution of the sample shows that 35.2% of the respondents are in their 50s and 37.1% of the respondents are in their 60s. The education level of older adults show that 35.1% of respondents are illiterate and 18.9% do not have formal schooling but can read and/or write. As for communities, the sample has 40.1% of urban communities, which means 59.9% of rural communities. The mean score of infrastructure sufficiency index of a community is 7.77. There are 13.7% of communities with older population higher than 25% and 26.1% of communities with college educated residents higher than 5% of the total residents.

Table 1.

Descriptive Statistics.

	M (SD)/percentage
Individual characteristics (n = 6,290)
Paid work	0.084 (0.277)
Grandparental childcare	0.220 (0.414)
Community, leisure, and social activities	0.226 (0.418)
Age	64.402 (9.307)
Female	0.539 (0.499)
Has a partner	0.754 (0.431)
Education level
Illiterate	35.12%
Capable of reading and/or writing; elementary	40.85%
Middle school and above	24.04%
Household expenditure per capita	7.21 (1.35)
At least one ADL disability	0.079 (0.269)
Receives pension	0.131 (0.338)
Live with at least one child	0.327 (0.469)
Hukou changed	0.453 (0.498)
Community characteristics (n = 307)
Urban communities	0.401 (0.491)
Infrastructure sufficiency index	7.77 (5.23)
Ratio of public facilities	0.577 (0.704)
Percentage of older population	0.160 (0.148)
Percentage of college-educated residents
0-1	35.18
1-5	40.72
Above 5	24.10
Percentage of migrant population	0.66 (0.139)

Note. ADL = activities of daily living.

Community Environments and Paid Work

Before interpreting the results (Table 2), we ran an “empty” model to check the validity of specifying a two-level model by including the dependent variable and the result showed that there is significant variations due to community differences (results not shown). Thus, it has statistical support to specify a two-level model than specifying a single-level model. The ICC in Model 4 declines to 13.7%. The changes show that adding variables from the three dimensions of community environment explains the variances.

Table 2.

Multilevel Logistic Models of Paid Work (Unweighted).

	Model 1	Model 2	Model 3	Model 4
	Odds ratio (SE)	Odds ratio (SE)	Odds ratio (SE)	Odds ratio (SE)
Individual characteristics
Household expenditure per capita	1.084* (0.041)	1.059 (0.039)	1.034 (0.038)	1.032 (0.038)
Receives pension	1.167 (0.166)	1.128 (0.158)	1.069 (0.148)	1.068 (0.147)
Female	0.405*** (0.047)	0.387*** (0.044)	0.382*** (0.043)	0.382*** (0.043)
Age groups (reference = 50-59) years
60-69	0.480*** (0.049)	0.462*** (0.047)	0.447*** (0.045)	0.446*** (0.045)
70+	0.151*** (0.027)	0.139*** (0.025)	0.130*** (0.023)	0.129*** (0.023)
Has a partner	0.908 (0.126)	0.907 (0.125)	0.916 (0.126)	0.908 (0.125)
Education levels (reference = illiterate)
Capable of reading and/or writing	1.112 (0.172)	1.037 (0.161)	0.992 (0.153)	0.994 (0.153)
Elementary/home school	1.284^† (0.192)	1.152 (0.172)	1.080 (0.160)	1.082 (0.161)
Middle school or above	1.580** (0.228)	1.316^† (0.190)	1.185 (0.170)	1.190 (0.172)
At least one ADL disability	0.475** (0.132)	0.484** (0.134)	0.505* (0.139)	0.507* (0.140)
Lives with at least one child	0.775* (0.082)	0.764* (0.080)	0.749** (0.078)	0.742** (0.077)
Hukou changed	1.181 (0.128)	1.121 (0.120)	1.075 (0.114)	1.069 (0.113)
Community characteristics
Urban communities		3.045*** (0.413)	1.528** (0.225)	1.423* (0.215)
Infrastructure sufficiency index			1.120*** (0.017)	1.112*** (0.017)
Ratio of public facilities (reference = 0-0.41)
>0.41			0.871 (0.102)	0.868 (0.101)
Percentage of older population (reference = 0-11)
12-24				0.883 (0.113)
Above 25				0.958 (0.179)
Percentage of migrant population (reference = 0)
0-5				1.178 (0.176)
Above 5				1.421* (0.219)
Percentage of college-educated residents (reference = 0-1)
1-5				1.134 (0.158)
Above 5				1.136 (0.184)
Constant	0.112 (0.038)	0.105***(0.035)	0.084*** (0.028)	0.079 (0.028)
$σ_{u}^{2}$	0.963 (0.159)	0.673 (0.124)	0.413 (0.094)	0.384 (0.09)
$ρ_{u}$	0.277	0.211	0.141	0.137
Observations	6,290	6,290	6,290	6,290

Note. Standard error in parentheses. ADL = activities of daily living; CHARLS = China Health and Retirement Longitudinal Study.

Source. Own calculations using CHARLS Wave 1.

†

<.1. *p < .05. **p < .01. ***p < .001.

From Model 2 to Model 4, older adults living in urban communities are linked to a higher likelihood of working for pay, as the odds ratios (ORs) are significantly positive. This supports the first hypothesis (H1): Older adults in urban areas are more likely to engage in the activities (paid work), whereas older adults in rural communities are less likely to do so. Moreover, the ORs for an urban community decrease as the variables of institutional and sociodemographic environments are added (from OR = 3.045 in Model 2 to OR = 1.423 in Model 4). It shows that the variables from the other two community environments have independent effects on the likelihood of engaging in paid work. A community with a higher score in infrastructure sufficiency index is associated with higher likelihood of paid work (OR = 1.120, p < .001), supporting the second hypothesis, (H2): Older adults in communities with better community infrastructure are more likely to engage in the activities (paid work) than those in communities with poorer community infrastructure.

For the sociodemographic environment of a community, it is found that a community with a higher percentage of migrant population is associated with engagement in paid work (OR = 1.421, p < .05). This is opposite to H6: older adults in communities with higher percentage of migrant population are less likely to participate in the activities (paid work).³

Community Environments and Grandparental Childcare

Before interpreting the results (Table 2), we ran a model to check the validity of specifying a two-level model and the result showed that there is significant variations due to community differences (results not shown). In Table 3, the ICC at the community level (ρ) decreases as community variables are added from Model 2 to Model 4, indicating that adding the community variables reduces the between-community variances to a certain degree. And, adding all community variables in Model 4 has improved the model fit (likelihood tests, results not shown).

Table 3.

Multilevel Logistic Models of Grandparental Childcare (Unweighted).

	Model 1	Model 2	Model 3	Model 4
	Odds ratio (SE)	Odds ratio (SE)	Odds ratio (SE)	Odds ratio (SE)
Individual characteristics
Household expenditure per capita	0.891*** (0.021)	0.883*** (0.021)	0.882*** (0.021)	0.881*** (0.021)
Receives pension	0.956 (0.091)	0.933 (0.089)	0.930 (0.089)	0.931 (0.090)
Female	1.002 (0.074)	0.981 (0.072)	0.980 (0.072)	0.981 (0.072)
Age groups (reference = 50-59) years
60-69	0.834** (0.055)	0.829** (0.055)	0.831** (0.055)	0.825** (0.054)
70+	0.231*** (0.024)	0.225*** (0.023)	0.225*** (0.023)	0.225*** (0.023)
Has a partner	1.059 (0.085)	1.056 (0.084)	1.056 (0.084)	1.050 (0.084)
Education levels (reference = illiterate)
Capable of reading and/or writing	1.424*** (0.124)	1.392*** (0.122)	1.386*** (0.122)	1.391*** (0.122)
Elementary/home school	1.201* (0.107)	1.158 (0.104)	1.154 (0.104)	1.157 (0.104)
Middle school or above	1.340** (0.120)	1.261* (0.115)	1.254* (0.115)	1.266* (0.116)
At least one ADL disability	0.649** (0.090)	0.652** (0.090)	0.653** (0.091)	0.649** (0.090)
Lives with at least one child	4.343*** (0.283)	4.306*** (0.280)	4.306*** (0.280)	4.334*** (0.283)
Hukou changed	1.070 (0.074)	1.062 (0.073)	1.060 (0.073)	1.057 (0.073)
Community environments
Urban communities		1.310*** (0.105)	1.271* (0.125)	1.300** (0.130)
Infrastructure sufficiency index			1.004 (0.010)	1.007 (0.010)
Ratio of public facilities (reference = 0-0.41)
>0.41			1.103 (0.082)	1.078 (0.080)
Percentage of older population (reference = 0-11)
12-25				0.981 (0.080)
Above 25				1.275* (0.145)
Percentage of migrant population (reference = 0)
0-5				0.864 (0.080)
Above 5				0.994 (0.095)
Percentage of college-educated residents (reference = 0-1)
1-5				1.187* (0.101)
Above 5				1.099 (0.119)
Constant	0.463*** (0.096)	0.474*** (0.098)	0.446*** (0.094)	0.413*** (0.091)
$σ_{u}^{2}$	0.199 (0.041)	0.186 (0.039)	0.182 (0.039)	0.166 (0.038)
$ρ_{u}$	0.079	0.073	0.072	0.066
Observations	6,290	6,290	6,290	6,290

Note. Standard error in parentheses. ADL = activities of daily living. CHARLS = China Health and Retirement Longitudinal Study.

Source. Own calculations using CHARLS Wave 1.

†

<.1. *p < .05. **p < .01. ***p < .001.

Living in urban communities is positively associated with engagement in grandparental childcare (OR = 1.300, p < .01). H1 is supported: Older adults in urban areas are more likely to engage in activities (grandparental childcare) whereas older adults in rural communities are less likely to do so. Older adults living in a community with higher ratio of public facilities are more likely to provide grandparental childcare (OR = 1.275, p < .05). H3 is supported. Another variable that is associated with higher likelihood of grandparental childcare is percentage of college-educated residents. Hence, older adults in a community with higher percentage of college-educated residents are more likely to participate in activities (grandparental childcare; H4). Older adults in a community with a higher percentage of older population (25% above) are more likely to engage in grandparental childcare (OR = 1.275, p < .05). This is opposite to H5: older adults are less likely to engage in activities (grandparental childcare) in communities in which a higher percentage of the population is older. An additional analysis reveals that there are interaction effects between urban communities and the categories of the percentage of older population (results not shown).

Community Environments and Community and Leisure Activities

We also checked that specifying a two-level model is better than specifying a single-level model as there are statistically significant community variations. Decreasing ICC is observed in Table 4, which means that variations between communities are due to the indicators of community environment.

Table 4.

Multilevel Logistic Models of Community and Leisure Activities (Unweighted).

	Model 1	Model 2	Model 3	Model 4
	Odds ratio (SE)	Odds ratio (SE)	Odds ratio (SE)	Odds ratio (SE)
Individual characteristics
Household expenditure per capita	1.076** (0.028)	1.059* (0.027)	1.051^† (0.027)	1.048^† (0.027)
Receives pension	1.433*** (0.137)	1.394*** (0.132)	1.361** (0.129)	1.362** (0.129)
Female	0.669*** (0.052)	0.648*** (0.051)	0.643*** (0.050)	0.644*** (0.050)
Age groups (reference = 50-59) years
60-69	0.932 (0.070)	0.919 (0.069)	0.914 (0.069)	0.910 (0.068)
70+	0.758** (0.073)	0.721*** (0.069)	0.707*** (0.068)	0.702*** (0.068)
Has a partner	0.953 (0.084)	0.951 (0.084)	0.956 (0.084)	0.953 (0.084)
Education levels (reference = illiterate)
Capable of reading and/or writing	1.549*** (0.156)	1.498*** (0.151)	1.473*** (0.149)	1.471*** (0.148)
Elementary/home school	1.857*** (0.182)	1.761*** (0.173)	1.717*** (0.169)	1.719*** (0.169)
Middle schools or above	2.444*** (0.243)	2.224*** (0.222)	2.143*** (0.215)	2.144*** (0.215)
At least one ADL disability	0.504*** (0.074)	0.511*** (0.075)	0.519*** (0.076)	0.519*** (0.076)
Lives with at least one child	1.063 (0.077)	1.052 (0.076)	1.047 (0.076)	1.042 (0.075)
Hukou changed	1.324*** (0.099)	1.298*** (0.097)	1.282*** (0.096)	1.275** (0.095)
Community environments
Urban communities		1.787*** (0.177)	1.311* (0.161)	1.179 (0.145)
Infrastructure sufficiency index			1.049*** (0.012)	1.036** (0.013)
Ratio of public facilities (reference = 0-0.41)
>0.41			1.038 (0.096)	0.925 (0.092)
Percentage of older population (reference = 0-11)
12-25				1.090 (0.152)
Above 25				1.131 (0.131)
Percentage of migrant population (reference = 0)
0-5				1.385** (0.164)
Above 5				1.288* (0.137)
Percentage of college-educated residents (reference = 0-1)
1-5				1.288* (0.137)
Above 5				1.597*** (0.209)
Constant	0.121*** (0.028)	0.119*** (0.028)	0.100*** (0.024)	0.087*** (0.022)
$σ_{u}^{2}$	0.483 (0.074)	0.427 (0.068)	0.397 (0.06)	0.352 (0.060)
$ρ_{u}$	0.138	0.124	0.116	0.104
Observations	6,290	6,290	6,290	6,290

Note. Standard error in parentheses. ADL = activities of daily living. CHARLS = China Health and Retirement Longitudinal Study.

Source. Own calculations using CHARLS Wave 1.

†

<.1. *p < .05. **p < .01. ***p < .001.

First of all, the effect of urban communities becomes insignificant as variables from the institutional and sociodemographic environments are included (Model 4). It suggests that the effect of the economic environment is explained by the institutional and sociodemographic environment. Model 4 indicates that older adults are more likely to engage in community and leisure activities in a community with a higher score in the sufficient infrastructure index than those with a lower score in the index (OR = 1.036, p < .01). This supports H2: Older adults in communities with better community infrastructure are more likely to engage in activities (community and leisure activities) than those in communities with poorer community infrastructure. Furthermore, the percentages of college-educated residents are positively associated with the likelihood of social and leisure activities (1%-5%, OR = 1.288, p < .05; >5%, OR = 1.597, p < .001). H4 is supported: Older adults in a community with higher percentages of college-educated residents are more likely to participate in activities (community and leisure activities) than those in a community with no college-educated residents. Lastly, older adults in a community with a higher percentage of migrant population (>5%) will be more likely to participate in community and leisure activities than those in a community with no migrant population (OR = 1.385, p < .01). Again, this is opposite to H6.

Additional Analyses

Appendix Table A1, A2, and A3 demonstrate the multilevel models of paid work, grandparental childcare, community activities and leisure activities based on the weighted samples. The findings are similar to the findings from the unweighted sample, but the significance of urban/rural communities (on the likelihood of paid work), the percentage of older population and the percentage of college-educated residents (on grandparental care) reduced to the margins. It gives us a conservative interpretation in these indicators later.

We include agricultural activities in the additional analyses (Appendix Table A4). The results show the differences in the individual characteristics in comparison with paid work. Those who are poorer, lower educated and married are more likely to engage in the agricultural activities. Older adults who live in communities with poorer economic environment, lacking of institutional resources, fewer migrants, and fewer college graduates are more likely to engage in such activities.

Discussion and Conclusion

To understand community factors of Active Aging in China, we develop three conceptual dimensions of community environments—the economic, institutional and sociodemographic environments. Against the background of building long-term care facilities and programs for China’s rapid aging population, our study contributes to the body of long-term care literature as we examine relevant community environment that help older adults extend their active profiles before they have to use long-term care resources.

The significance of the economic environment to engagement in paid work suggests that better economic development provides more job opportunities for older adults but the impacts decline after including indicators of institutional environment However, for grandparental childcare, community activities and leisure activities, the impact of economic development disappears when the indicators from the institutional and sociodemographic environment are added. On one hand, the result is in line with the study by Feng et al. (2011), which found that urban communities have greater resources and services for their older residents. On the other hand, our results suggest that urban–rural differences actually reflect the gap of infrastructure establishment, which is essential for older Chinse in participating activities. The institutional environment, especially community infrastructure sufficiency, is a relevant indicator to engagement in paid work, community activities and leisure activities. It implies an intertwined relationship between the economic and institutional environment of a community. The implication reinforces the point that the government as the third party should build up community capacities in supporting older adults’ activity engagement (Haski-Leventhal et al., 2010). We suggest that further research integrate with the policy framing to understand how social policies in China promote active engagement.

Percentage of college-educated residents are found to be relevant to paid work, and community activities and leisure activities. The percentage of college-educated residents are associated with higher likelihood of grandparental childcare, community activities and leisure activities, while controlling for the economic development of a community. Okun and Michel (2006) and Wilson (2000) explain that residents with higher education would be more involved in social activities, especially volunteering. Our findings on community activities and leisure activities support such an explanation.

The percentage of migrant population is positively related to paid work and community activities and leisure activities. Although the literature mentions that feeling secure and stable in their communities is relevant to older adults’ engagement (Dury et al., 2014; Okun & Michel, 2006), our findings do not support such argument. A community with a higher percentage of migrant population may provide more job opportunities for older adults to engage in economic activities as labor migration is closely related to its economic development. For community and leisure activities, a community with a higher percentage of migrant population may facilitate social space and occasions for people to meet up. Such a social context indirectly stimulates atmosphere of activity engagement for older adults to join community and leisure activities. We call for further investigations to analyze the older residents’ perceptions of their residential communities with higher percentages of migrant population.

Although engagement in agricultural activities is not mentioned in literature on Active Aging, studies have documented that the prevalence of agricultural activities among older Chinese (H. Li, Chi, & Xu, 2013 on rural China; Y. C. Chen et al., 2018). Our additional analyses imply that older adults in disadvantaged community environments have higher likelihood to engage in such activities. Such results show that participation in agricultural activities are associated with the developing economy in China. It suggests that promoting Active Aging and designing long-term care programs shall consider the fact that majority of older adults relying on agricultural activities for a living. More institutional investment on improving infrastructure and facilities would be relevant to support older adults who are still involved in agricultural work. Improving the community capacities and resources will enhance the possibilities of planning Active Aging.

We acknowledge three limitations in this study. Firstly, the activities we included in the analyses did not capture the complete spectrum of activities in later life, nor could we look into the actual content of activities (e.g., what was involved in the volunteer work; Morrow-Howell, 2010; Morrow-Howell et al., 2014). If future data permits, it is necessary to consider the activity content in the definitions. In addition, aging actively for older adults may be understood differently from culture to culture (Hustinx, Cnaan, & Handy, 2010; Morrow-Howell & Wang, 2013). Further research may consider to investigate the meaning of Active Aging from the perspective of older adults to enrich its conceptualization. Secondly, this study analyzed the cross-sectional data and hence, it should be kept in mind that relationships are associations, rather than causal relationships. Lastly, the CHARLS did not include information of ethnicity of respondents. We were unable to consider the variations of older adults with different ethnic backgrounds.

To conclude, our study brings in knowledge of what kinds of community environments will create the opportunity structures for Active Aging, which provides insights on preparations for community-based long-term care in China. The role of institutional settings and sociodemographic characteristics of a community shows the need to go beyond the urban–rural divide and individual characteristics. Promoting activity engagement requires investment on community infrastructure and an understanding of the sociodemographic composition of older people’s residential communities. Our findings also suggest that community environments would be crucial for the execution of long-term care in China. We would like to call for continuous investigations.

Footnotes

Appendix

Table A4.

Multilevel Logistic Models of Agricultural Activities (Without Weights).

	Model 1	Model 2	Model 3	Model 4
	Odds ratio (SE)	Odds ratio (SE)	Odds ratio (SE)	Odds ratio (SE)
Individual characteristics
Household expenditure per capita	0.860*** (0.021)	0.870*** (0.021)	0.877*** (0.021)	0.880*** (0.021)
Receives pension	0.805* (0.083)	0.826 (0.084)	0.864 (0.087)	0.871 (0.088)
Female	0.700*** (0.054)	0.716*** (0.055)	0.716*** (0.055)	0.714*** (0.055)
Age groups (reference = 50-59) years
60-69	0.725*** (0.054)	0.735*** (0.054)	0.742*** (0.055)	0.740*** (0.055)
70+	0.204*** (0.019)	0.214*** (0.020)	0.221*** (0.020)	0.219*** (0.020)
Has a partner	2.206*** (0.180)	2.231*** (0.181)	2.220*** (0.180)	2.206*** (0.179)
Education levels (reference = illiterate)
Capable of reading and/or writing	1.001 (0.092)	1.040 (0.095)	1.076 (0.098)	1.082 (0.098)
Elementary/home school	0.932 (0.085)	0.978 (0.089)	1.022 (0.093)	1.026 (0.093)
Middle schools or above	0.585*** (0.059)	0.635*** (0.063)	0.667*** (0.066)	0.674*** (0.066)
At least one ADL disability	0.246*** (0.031)	0.245*** (0.031)	0.239*** (0.030)	0.238*** (0.030)
Lives with at least one child	1.121 (0.079)	1.135 (0.080)	1.143 (0.080)	1.154* (0.081)
Hukou changed	0.851* (0.062)	0.869 (0.063)	0.894 (0.065)	0.901 (0.065)
Community environments
Urban communities		0.129*** (0.018)	0.352*** (0.049)	0.418*** (0.058)
Infrastructure sufficiency index			0.860*** (0.012)	0.876*** (0.012)
Ratio of public facilities (reference = 0-0.41)
>0.41			1.149 (0.124)	1.112 (0.118)
Percentage of older population (reference = 0-11)
12-25				1.247 (0.144)
Above 25				1.292 (0.207)
Percentage of migrant population (reference = 0)
0-5				0.701** (0.091)
Above 5				0.904 (0.122)
Percentage of college-educated residents (reference = 0-1)
1-5				1.137 (0.135)
Above 5				0.608** (0.096)
$σ_{u}^{2}$	1.97	1.00	0.657	0.596
$ρ_{u}$	0.419	0.268	0.194	0.179
Observations	6,289	6,289	6,289	6,289

Note. Standard error in parentheses. ADL = activities of daily living. CHARLS = China Health and Retirement Longitudinal Study.

Source. Own calculations using CHARLS Wave 1.

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

Declaration of Conflicting Interests

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

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Notes

ORCID iD

Pei-Chun Ko

References

Annear

Keeling

Wilkinson

Cushman

Gidlow

Hopkins

(2014). Environmental influences on healthy and active ageing: A systematic review. Ageing & Society, 34, 590-622.

Avramov

Maskova

(2003). Population Studies Series No, 41. Active ageing in Europe, volume 1. Strasbourg, France: Council of Europe.

Bray

(2006). Building “community”: New strategies of governance in urban China. Economy and Society, 35, 530-549.

Chen

Coulson

N. E.

(2002). Determinants of urban migration: Evidence from Chinese cities. Urban Studies, 39, 2189-2197.

Chen

Y. C.

Wang

Cooper

McBride

Chen

Wang

. . . Morrow-Howell

(2018). A research note on challenges of cross-national aging research: An example of productive activities across three countries. Research on Aging, 40, 54-71.

Day

(2008). Local environments and older people’s health: Dimensions from a comparative qualitative study in Scotland. Health & Place, 14, 299-312.

Yang

(2010). China’s population ageing and active ageing. China Journal of Social Work, 3, 139-152.

Dury

Willems

De Witte

De Donder

Buffel

Verté

(2014). Municipality and neighborhood influences on volunteering in later life. Journal of Applied Gerontology, 35, 601-626.

Feng

Zhan

H. J.

Feng

Liu

Sun

Mor

(2011). An industry in the making: The emergence of institutional elder care in urban China. Journal of the American Geriatrics Society, 59, 738-744.

10.

Gessa

G. D.

Grundy

(2014). The relationship between active ageing and health using longitudinal data from Denmark, France, Italy and England. Journal of Epidemiology & Community Health, 68(3), 261-267.

11.

Glass

T. A.

Balfour

J. L.

(2003). Neighborhoods, aging, and functional limitations. In Kawachi

Berkman

L. F.

(Eds.), Neighborhoods and health (pp. 303-334). New York, NY: Oxford University Press.

12.

Guan

Chow

(2003). The development of urban community welfare services in China. Social Development Issues, 25(3), 51-61.

13.

Haski-Leventhal

Meijs

L. C.

Hustinx

(2010). The third-party model: Enhancing volunteering through governments, corporations and educational institutes. Journal of Social Policy, 39, 139-158.

14.

Henkin

Zapf

(2006-2007). How communities can promote civic engagement of people age 50-plus. Generations, 30, 72-77.

15.

Herzog

A. R.

Morgan

J. N.

(1992). Age and gender differences in the value of productive activities: Four different approaches. Research on Aging, 14, 169-198.

16.

Hussein

Manthorpe

(2014). Structural marginalisation among the long-term care workforce in England: Evidence from mixed-effect models of national pay data. Ageing & Society, 34, 21-41.

17.

Hustinx

Cnaan

R. A.

Handy

(2010). Navigating theories of volunteering: A hybrid map for a complex phenomenon. Journal for the Theory of Social Behaviour, 40, 410-434.

18.

Lawton

M. P.

(1982). Competence, environmental press, and the adaptation of older people. In Lawton

Windley

Byerts

T. O.

(Eds.), Aging and the environment: Theoretical approaches (pp. 33-59). New York, NY: Springer.

19.

Lei

Sun

Strauss

Zhang

Zhao

(2014). Depressive symptoms and SES among the mid-aged and elderly in China: Evidence from the China Health and Retirement Longitudinal Study national baseline. Social Science & Medicine, 120, 224-232.

20.

Chi

(2010). Factors associated with volunteerism among community-living older adults in urban China. Social Development Issues, 32(1), 62-75.

21.

Chi

(2013). Life satisfaction of older Chinese adults living in rural communities. Journal of Cross-Cultural Gerontology, 28, 153-165.

22.

L. W.

Liu

Zhang

(2016). Understanding rural–urban differences in depressive symptoms among older adults in China. Journal of Aging and Health, 28, 341-362.

23.

Zhang

Zhou

Chen

(2013). Rural-urban differences in the long-term care of the disabled elderly in China. PLoS ONE, 8(11), e79955.

24.

Liu

Lou

W. Q.

(2016). Patterns of productive activity engagement among older adults in urban China. European Journal of Ageing, 13, 361-372.

25.

Liu

Sun

(2016). Pension reform in China. Journal of Aging & Social Policy, 28, 15-28.

26.

Lum

T. Y. S.

(2013). Advancing research on productive aging activities in greater Chinese societies. Ageing International, 38, 171-178.

27.

Morrow-Howell

(2010). Volunteering in later life: Research frontiers. The Journals of Gerontology, Series B: Psychological Sciences & Social Sciences, 65, 461-469.

28.

Morrow-Howell

Putnam

Lee

Y. S.

Greenfield

J. C.

Inoue

Chen

(2014). An investigation of activity profiles of older adults. The Journals of Gerontology, Series B: Psychological Sciences & Social Sciences, 69, 809-821.

29.

Morrow-Howell

Wang

(2013). Productive engagement of older adults: Elements of a cross-cultural research agenda. Ageing International, 38, 159-170.

30.

Musick

Wilson

(2008). Volunteers: A social profile. Indianapolis: Indiana University Press.

31.

Okun

M. A.

Michel

(2006). Sense of community and being a volunteer among the young-old. Journal of Applied Gerontology, 25, 173-188.

32.

Paillard-Borg

Wang

H.-X.

Winblad

Fratiglioni

(2009). Pattern of participation in leisure activities among older people in relation to their health conditions and contextual factors: A survey in a Swedish urban area. Ageing & Society, 29, 803-821.

33.

Pang

De Brauw

Rozelle

(2004). Working until you drop: The elderly of rural China. The China Journal, 52, 73-94.

34.

Park

(2008). Rural-urban inequality in China. In Yusufy

Saich

(Eds.), China urbanizes: Consequences, strategies and policies, East Asia economic prospects (pp. 42-63). Washington DC, USA: The World Bank.

35.

Richard

Gauvin

Gosselin

Laforest

(2009). Staying connected: Neighbourhood correlates of social participation among older adults living in an urban environment in Montreal, Quebec. Health Promotion International, 24, 46-57.

36.

Riley

M. W.

Riley

J. W.

Jr. (1994). Structural lag: Past and future. Oxford, UK: John Wiley.

37.

Rodrigues

Hofmarcher

M. M.

Leichsenring

Winkelmann

(2013). Active and healthy ageing for better long-term care: A fresh look at innovative practice examples (European Programme for Employment and Social Solidarity Progress). Retrieved from file:///G:/ProductiveAgingProject/Topic_EcologyofActiveAging/Fulltext/Reviews_JAH/CZ_Active_Ageing_Policy_final_revised.pdf

38.

Rotolo

Wilson

(2012). State-level differences in volunteerism in the United States: Research based on demographic, institutional, and cultural macrolevel theories. Nonprofit and Voluntary Sector Quarterly, 41, 452-473.

39.

Rowles

G. D.

Bernard

M. A.

(2013). Environmental gerontology: Making meaningful places in old age. New York, NY: Springer.

40.

Saich

(2000). Negotiating the state: The development of social organizations in China. The China Quarterly, 161, 124-141.

41.

Shen

(2014). Community building and mental health in mid-life and older life: Evidence from China. Social Science & Medicine, 107, 209-216.

42.

State Council. (2001). The 10th five year plan for the development of China’s undertakings for the aged. Retrieved from http://www.gov.cn/gongbao/content/2001/content_60985.htm (in Chinese)

43.

Swaroop

Morenoff

J. D.

(2006). Building community: The neighborhood context of social organization. Social Forces, 84, 1665-1695.

44.

Wahl

H.-W.

Weisman

G. D.

(2003). Environmental gerontology at the beginning of the new millennium: Reflections on its historical, empirical, and theoretical development. The Gerontologist, 43, 616-627.

45.

Wilson

(2000). Volunteering. Annual Review of Sociology, 26, 215-240.

46.

World Health Organization. (2002). Active ageing: A policy framework. Geneva, Switzerland: Author. Retrieved from http://apps.who.int/iris/bitstream/10665/67215/1/WHO_NMH_NPH_02.8.pdf

47.

Carter

M. W.

Goins

R. T.

Cheng

(2005). Emerging services for community-based long-term care in urban China: A systematic analysis of Shanghai’s community-based agencies. Journal of Aging & Social Policy, 17(4), 37-60.

48.

(2007). Community participation in urban China: Identifying mobilization factors. Nonprofit and Voluntary Sector Quarterly, 36, 622-642.

49.

Chow

J. C.

(2006). Urban community in China: Service, participation and development. International Journal of Social Welfare, 15, 199-208.

50.

Yan

M. C.

Gao

J. G.

(2007). Social engineering of community building: Examination of policy process and characteristics of community construction in China. Community Development Journal, 42, 222-236.

51.

Yeatts

D. E.

Cready

C. M.

Pei

Shen

Luo

(2014). Environment and subjective well-being of rural Chinese elderly: A multilevel analysis. The Journals of Gerontology, Series B: Psychological Sciences & Social Sciences, 69, 979-989.

52.

Yeatts

D. E.

Pei

Cready

C. M.

Shen

Luo

Tan

(2013). Village characteristics and health of rural Chinese older adults: Examining the CHARLS Pilot Study of a rich and poor province. Social Science & Medicine, 98, 71-78.

53.

Zhang

N. J.

Guo

Zheng

(2012). China: Awakening giant developing solutions to population aging. The Gerontologist, 52, 589-596.

54.

Zhang

Fan

Zhang

Huang

(2004). Local governance and public goods provision in rural China. Journal of Public Economics, 88, 2857-2871.

55.

Zhao

Strauss

Yang

Giles

P. P.

. . . Wang

(2013). China Health and Retirement Longitudinal Study—2011-2012 national baseline users’ guide. Retrieved from http://charls.pku.edu.cn/uploads/document/2011-charls-wave1/application/CHARLS_nationalbaseline_users_guide.pdf

56.

Zimmer

Kaneda

Spess

(2007). An examination of urban versus rural mortality in China using community and individual data. The Journals of Gerontology, Series B: Psychological Sciences & Social Sciences, 62, S349-S357.

57.

Zimmer

Kwong

(2003). Family size and support of older adults in urban and rural China: Current effects and future implications. Demography, 40, 23-44.

58.

Zimmer

Wen

Kaneda

(2010). A multi-level analysis of urban/rural and socioeconomic differences in functional health status transition among older Chinese. Social Science & Medicine, 71, 559-567.