Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 279
12
Effects of Social Activities on
Cognitive Functions:
Evidence from CHARLS1
Yuqing Hu, Xiaoyan Lei, James P. Smith, and Yaohui Zhao
C
ognitive function is a key dimension of the quality of life for the
elderly in all countries. It is closely related to the ability to process
information in daily life and helps shape overall well-being over
the life course (McArdle, Smith, and Willis, 2011). Cognitive function
involves operations such as perception, memory, creation of imagery, and
thinking, and it declines sharply as people approach advanced age (Levy,
1994; Tilvis et al., 2004). With population aging, more people will begin
to suffer from cognitive impairment. The situation is especially severe in
China because of its extraordinary speed of population aging, accompa -
nied by the eventual decline of families as a source of eldercare and the
lack of long-term care facilities. It is thus important to study determinants
of cognitive function to understand how to best postpone and slow down
its eventual decline.
Social activities are of particular interest to us as they play a signifi-
cant role in the daily lives of most Chinese elderly. Social-emotional Selec-
1 This research was supported by grants from the National Institute on Aging to CCER
at Peking University and the RAND Corporation, and grants from the Natural Science
Foundation of China. It was also supported by the National Institute on Aging (Grant
Number R21AG031372), Natural Science Foundation of China (Grant Numbers 70773002
and 70910107022), the World Bank (Contract 7145915), and the Fogarty International Center
(Grant Number R03TW008358). The content is solely the responsibility of the authors and
does not necessarily represent the official views of any of the funders. We appreciate the
helpful comments received at the National Academy of Sciences meeting in New Delhi,
India, and from the NAS reviewers.
279
OCR for page 280
280 AGING IN ASIA
tivity Theory (SST) argues that as people age, they become increasingly
selective by investing greater resources in emotionally meaningful goals
and activities. Present-oriented goals with emotional meaning are pri -
oritized over future-oriented goals aimed at acquiring information about
future decisions and expanding horizons (Löckenhoff and Carstensen,
2004). Due to a gradual decline in cognitive functioning and the ability to
deal with complex cognitive tasks, the elderly seek out more emotional
goals. Such motivations further limit their ability to seek information
as well as their attention, memory, and cognition processing. However,
cognitively stimulating social activity is hypothesized to benefit cognitive
functions by providing resistance to mental diseases, such as dementia,
and by reducing rates of cognitive decline (Hsu, 2007; Wang et al., 2002).
The effects of different social activities on cognitive functions of the old
warrant more exploration, especially for China where little research cur-
rently exists in part due to the lack of high-quality micro data.
A large volume of literature has investigated the relationship between
cognition and social activities or social engagement among the elderly
(Allaire and Marsiske, 1999; Wang et al., 2002; Zunzunegui et al., 2003).
These studies confirm a positive relationship between the two. Among
these studies, Zunzunegui et al. (2003), using data from a longitudinal
survey of community-dwelling people over age 65, analyzed causal
effects of social networks, social integration, and social engagement on
cognitive decline of community-dwelling older Spanish adults with
social variables measured at baseline, and cognitive change and decline
measured after four years of follow-up. They were unable to determine
whether the observed effect of social relations on cognitive function was
the result of cumulative lifelong exposure to extensive social networks or
a consequence of an abrupt change from an extensive network to a more
limited one.
Similar studies on the relation between cognition and social activities
in developing countries such as China are more fragmentary, primarily
due to the lack of relevant data and a concentration on the young rela -
tive to health of the elderly. Asian populations are of particular interest
because their elderly people are more likely to reside with their children
and social activities may play different roles in their lives than in the west-
ern world. Glei et al. (2005) and Hsu (2007) are two exceptions that used
longitudinal data from a survey of elderly in Taiwan and explored the
effects of social participation on cognitive function. Hsu (2007) focused on
regular social group participation and ignored leisure activity. He found
that participating in any of these social groups has no significant correla-
tion with cognitive function. Glei et al. (2005) used a broader definition
of social interaction that includes more recreational activities and found
that participating in these social activities may play an important role
OCR for page 281
281
YUQING HU, XIAOYAN LEI, JAMES P. SMITH, and YAOHUI ZHAO
on delaying cognitive decline. In our study, we employ a similar defini-
tion of social activity as in Glei et al. and investigate data from mainland
China, which has the same tradition of living arrangements but is at a
different stage of economic development under a different institutional
background than Taiwan. In addition to what has been done in Hsu (2007)
and Glei et al. (2005), we will try to seek a causal explanation through
exploring the community-level facilities and organizations that enable
more social interactions among Chinese elderly.
More specifically, this chapter attempts to address three questions.
First, what is the relationship between social activity and different dimen -
sions of cognitive function for Chinese elderly? Second, what are the differ-
ences among effects of alternative social activities on different dimensions
of cognitive abilities? Instead of directly identifying the causal relationship,
we test in an ordinary least squares (OLS) reduced form setting whether
community facilities, which are strongly correlated with social activities,
are correlated to cognitive function. To address these issues, we will use
the recently collected data from the 2008 pilot survey of the China Health
and Retirement Longitudinal Study (CHARLS). In addition to OLS esti-
mates of association, we present OLS reduced form estimates that include
some arguably exogenous determinants of individuals’ social activities.
The chapter is organized as follows. The next section presents a sum-
mary of CHARLS data, including a description of variables and summary
statistics. The following section shows the principal relationships between
cognitive functions and social activities emerging from the CHARLS data.
We then present OLS reduced form estimations showing the relationship
between cognitive functions and community facilities. The final section
highlights our main conclusions.
DATA, VARIABLES, AND SUMMARY STATISTICS
Data and Variables
The 2008 CHARLS pilot was conducted in Zhejiang and Gansu prov-
inces. Zhejiang, located in the developed coastal region, is a dynamic
province with fast economic growth, a private sector, small-scale indus -
trialization, and an export orientation. In contrast, Gansu, located in the
less developed western region, is one of the poorest, most rural provinces
in China. These two provinces were selected in part due to their economic
diversity. Among all provinces in 2008, Zhejiang had the highest rural and
urban incomes per capita after Shanghai and Beijing, while Gansu had
the second lowest rural per capita income and fourth lowest urban per
capita income. The target population of CHARLS is individuals aged 45
and older and their spouses/partners irrespective of age.
OCR for page 282
282 AGING IN ASIA
The sampling design of the 2008 wave of CHARLS aimed to be rep-
resentative of residents aged 45 and older in these two provinces. Within
each province, CHARLS randomly selected 13 county-level units by PPS
(Probability Proportional to Size), stratified by regions and urban/rural.
Within each county-level unit, CHARLS randomly selected three village-
level units (villages in rural areas and urban communities in urban areas)
by PPS as primary sampling units (PSUs). Within each PSU, CHARLS
randomly selected 25 dwellings in rural and 36 in urban areas from a com -
plete list of dwelling units generated from a mapping/listing operation. In
situations where more than one age-eligible household lived in a dwelling
unit, CHARLS randomly selected one. Within each household, one person
aged 45 and older was randomly chosen as the main respondent, and the
spouse was automatically included. Based on this sampling procedure,
one or two individuals in each household were interviewed depending
on marital status of the main respondent. The total sample size was 2,685
individuals in 1,570 households. The CHARLS pilot experience was very
positive. Overall response rate was 85%: 79% in urban areas and 90% in
rural areas. Response rates were about the same in the two provinces,
83.9% in Zhejiang and 85.8% in Gansu. The high response rates reflected
detailed procedures put in place to ensure a high response to the survey.2
In the CHARLS sample, people for the most part do not choose the
community in which they live, but instead they mostly live where they
were born. CHARLS respondents who are by design aged 45 and older
are relatively immobile and did not participate in the great Chinese migra-
tion patterns. About 9 in 10 CHARLS respondents are living in the same
county in which he or she was born and less than 1 in every 20 is living
in a different province in which he or she was born (Smith et al., 2012).
Following protocols of the Health and Retirement Studies (HRS)
international surveys, the 2008 CHARLS main questionnaire consists of
seven modules: (1) demographic background, (2) family, (3) health status
and functioning, (4) healthcare and insurance, (5) work, retirement and
pension, (6) income, expenditure and assets, and (7) interviewer observa -
tion (Zhao et al., 2009). All data were collected in face-to-face, computer-
aided personal interviews (CAPI).
Rich information makes CHARLS well suited for research on cogni-
tive abilities and social activities. Our study sample includes respondents
2 Letters to respondents were delivered (often by the village leader) to households to
inform them of the significance of the study, contents of questionnaires, provision of a
free physical examination, and compensation, as well as the expected date of arrival of the
interviewers. On the day of the interview, interviewers were often introduced to the house -
holds by community leaders to confirm the authenticity of their identity. When refused,
multiple attempts were made to persuade respondents to participate and the team leader
was required to go and make a final attempt before declaring the household a refusal.
OCR for page 283
283
YUQING HU, XIAOYAN LEI, JAMES P. SMITH, and YAOHUI ZHAO
45 years and older, representing Zhejiang and Gansu provinces. After
discarding 109 individuals younger than 45 years old (spouses of the main
respondents), 1 with missing age, and 282 with missing cognition due
to proxy,3 we are left with a sample of 2,293 observations, among which
1,131 are men and 1,162 are women, 850 (37.1%) are younger than 50 years
old, 768 (33.5%) are 50–60 years old, 461 (20.1%) are 60–70 years old, and
214 (9.3%) are aged 70 and older.
Social Activity (SA)
In CHARLS, questions on social activities are in the “health status
and function” module, where interviewees are asked whether they have
taken part in 10 specifically listed activities in the past month. The 10
activities are (1) volunteer or charity work; (2) caring for a sick or disabled
adult who does not live with the respondent and who does not pay for
the help; (3) providing help to family, friends, or neighbors who do not
live with the respondent and who do not pay for the help; (4) attending
an educational or training course; (5) interacting with friends; (6) playing
Mahjong, chess, or cards, or going to a community club; (7) attending
a sporting event or other kind of club; (8) taking part in a community-
related organization; (9) investing in stocks; and (10) surfing the Internet.
Seven out of these 10 are considered social activities, while the remaining
three activities—3, 9, and 10—are not.4
Table 12-1 depicts participation rates in each of these social activities.
Interacting with friends and playing table games such as Mahjong have
much higher participation rates than other activities (34.5% interacted
with friends and 17.1% played Mahjong, chess, or card games), so we put
each of them into a distinct group and grouped the rest into a category
called “other social activities.” We define a variable called “any activity”
to indicate if a respondent was involved in at least one social activity.
Forty-six percent participated in at least one activity.
Table games, especially the traditional Chinese game of Mahjong,
play an important role in many Chinese seniors’ lives. When relatives live
far away, Mahjong helps form a sense of belonging by providing a chance
to interact with those with whom one shares interests and personalities.
The effort to win games—remembering rules, learning skills, and observ-
ing and reacting to others’ behavior—forces the brain to work and is
hypothesized to enhance people’s cognitive functioning through such
3 Missing values of the control variables are imputed by multiple imputation method.
4 Options 9 and 10 are obviously not social activities. We do not consider option 3 as a
social activity, as the wording is very likely to be understood as providing economic help
to family and friends, which seems different from social activities in the Chinese context.
OCR for page 284
284 AGING IN ASIA
TABLE 12-1 Description of Social Activities and Cognitive Function
Gender
Social Activity and Other Activity Variables All Male Female
Interacted with friends 0.345 0.317 0.371
Played Mahjong, chess, or cards 0.171 0.236 0.108
Other social activities 0.086 0.076 0.095
Done voluntary or charity work 0.012 0.016 0.009
Cared for sick or disabled as volunteer 0.040 0.039 0.040
Attended education or training course 0.006 0.007 0.005
Gone to sport, social, or other club 0.027 0.014 0.039
Taken part in community-related organization 0.016 0.016 0.016
Attended any activity above 0.457 0.470 0.444
Number of activities attended 0.616 0.645 0.588
Number of observations 2,293 1,131 1,162
Cognitive function
Mental intactness (0-11) 8.621 9.046 8.167
[2.14] [1.89] [2.30]
Number of observations 1,873 968 905
Episodic memory 2.986 3.103 2.862
[2.02] [1.99] [2.05]
Number of observations 1,901 979 922
NOTE: For cognitive function, standard errors are below coefficients in brackets.
SOURCE: Data from CHARLS 2008 Pilot.
activities. Provided that sufficient infrastructure and services in the com -
munity exist, playing Mahjong or chess at home or in the neighborhood
is a common way for older Chinese to spend leisure time by interacting
with friends and neighbors.
Cognitive Ability (CA)
Cognitive function of respondents is also measured in the “health sta-
tus and function” module in the CHARLS questionnaire, which tests ori-
entation, calculation, word recall, and other cognitive dimensions. Cogni-
tive ability can be generally categorized into fluid cognitive ability (FCA)
and crystallized cognitive ability (CCA). The former concerns learning
OCR for page 285
285
YUQING HU, XIAOYAN LEI, JAMES P. SMITH, and YAOHUI ZHAO
Region Province Age Group
Urban Rural Zhejiang Gansu 70
0.385 0.313 0.405 0.271 0.400 0.311 0.323 0.290
0.229 0.125 0.230 0.101 0.195 0.175 0.141 0.131
0.129 0.051 0.104 0.064 0.118 0.073 0.067 0.042
0.020 0.006 0.011 0.014 0.019 0.009 0.004 0.014
0.047 0.034 0.041 0.039 0.059 0.038 0.024 0.005
0.011 0.002 0.007 0.005 0.011 0.004 0.002 0.005
0.056 0.003 0.039 0.012 0.037 0.023 0.020 0.014
0.020 0.013 0.022 0.008 0.022 0.009 0.017 0.009
0.535 0.394 0.556 0.338 0.514 0.419 0.451 0.374
0.767 0.496 0.756 0.448 0.742 0.569 0.532 0.467
1,017 1,276 1,253 1,040 850 768 461 214
9.153 8.140 9.144 7.979 9.032 8.593 8.156 7.687
[1.93] [2.21] [1.92] [2.23] [1.95] [2.13] [2.31] [2.26]
889 984 1032 841 741 648 353 131
3.411 2.608 3.083 2.869 3.498 3.022 2.388 1.544
[1.97] [1.99] [2.07] [1.96] [2.03] [1.97] [1.82] [1.63]
895 1,006 1,041 860 750 661 354 136
performance and processing of new material, which tends to decline
substantially in adulthood (Schaie, 1994; Verhaegen and Salthouse, 1997).
The latter includes knowledge and skills accumulated in the past, which
are not as easily lost.
In this research, for simplicity, we classify cognitive abilities into two
categories: mental intactness (MI) and episodic memory (EM). As shown
in Table 12-2, three items are used to measure mental intactness: a serial-7
number subtraction question, time orientation, and picture drawing. The
scores range from 0–5, 0–5, and 0–1, respectively. Orientation in time
consists of three questions about the interview date (day, month, year),
day of the week, and season, which were coded as dummies to indicate
whether answers are correct. To measure episodic memory, we use imme-
OCR for page 286
286 AGING IN ASIA
TABLE 12-2 Definition of Cognitive Abilities
Types Items Survey Questions Score
Mental Numerical What does 100 minus 7 equal? And 7 from that?.. 0~5
intactness ability
(0–11)
Time Please tell me today’s date (year, month, day). 0~3
orientation
Please tell me the day of the week. 0~1
What is the current season (among Spring, 0~1
Summer, Fall, or Winter)?
Picture Do you see this picture? Please draw that picture 0~1
drawing on this paper.
Episodic Immediate Try to remember the words I just read to you. 0~10
memory word recall I’ll ask you to recall them later.
(0–10)
Delayed A little while ago, I read you a list of words 0~10
word recall and you repeated the ones you could remember.
Please tell me any of the words that you
remember now.
NOTES: Scores have been adjusted so that higher values indicate better cognitive function
Episodic memory is the mean of scores of immediate and delayed word recalls.
SOURCE: Data from CHARLS 2008 Pilot.
diate and delayed word recall. In the recall test, respondents are read a list
of 10 simple nouns, then immediately asked to repeat as many of those
words as possible in any order. After 20 questions concerning CESD, they
are again asked to recall as many of the original words as possible. The
item is coded as 1 if recalled by the interviewee, and as 0 if not. Scores
for immediate and delayed recall both vary from 0 to 10. We follow the
approach by McArdle, Smith, and Willis (2011), using the mean of scores
in immediate and delayed word recall as the measure of episodic memory.
Figure 12-1 depicts the different downward trends of mental intactness
and episodic memory, where original average scores have been smoothed.
Clearly both cognitive measures steadily decline with age, with the age
pattern becoming more erratic as sample sizes in CHARLS 2008 become
thinner at older ages. One caution in reading this graph is that the number
of observations over aged 80 in the CHARLS data is relatively small (only
69 in total), so that the patterns start to become erratic at these ages.
Episodic memory is a very general measure of an important aspect of
fluid intelligence since access to memory is basic to any type of cognitive
ability. Most of the variation in this measure is picking out the low end—
people with bad memory so social activities of the sort analyzed here may
have noticeable impacts. Intact mental status, however, contains elements
of both fluid and crystallized intelligence, needed for many cognitive
tasks but not specific to any particular domain (see McArdle et al., 2011).
OCR for page 287
287
YUQING HU, XIAOYAN LEI, JAMES P. SMITH, and YAOHUI ZHAO
Average Score for Cognitive Function
Mental Intactness Episodic Memory
FIGURE 12-1 Cognition trend with age.
SOURCE: Data from CHARLS 2008 Pilot.
Other Variables
R02177
Other control variables are mainly concerned with demographics, socio-
economic status, and community-level12-1
Figure variables. Demographics include a
quadratic in age to capture any nonlinear effect of age, sex (female = 1),
bitmapped, uneditable, with labels and legend type replaced
marital status of the respondent (married = 1), location of household (urban
or rural [urbanfor 1]), province of residence (Zhejiang or Gansu), number
scaled = portrait, above, and for landscape, below
of children, and number of siblings. Number of children, including both
biological and non-biological daughters and sons, serves as one measure
for family support.
Socioeconomic status has two dimensions: education and log house -
hold expenditure per capita (log PCE). Education is classified into five dis-
Average Score for Cognitive Function
crete educational groups: illiterate, able to write or read, finished primary
school, finished junior high school, finished senior high school or above.
In the second category, “able to write or read” includes those not finish-
ing primary school but capable of reading or writing, or those reported to
have been in Sishu.5 The category of “high school or above” includes those
completing a senior high school, vocational school, college, or graduate
level education. We also control log PCE to capture effects of financial
resources. In rural developing economies, consumption expenditures rep-
resent the best measure of the economic resources available to the family
(Strauss and Thomas, 2008).
5 Sishu
is an old-style private Chinese education that mainly taught young children Chinese
classics before the 20th century.
Mental Intactness Episodic Memory
OCR for page 288
288 AGING IN ASIA
Lastly, we include community demographics and socioeconomic
status, since we believe that cognition may be affected by one’s surround-
ing environment (Engelhardt et al., 2010; Hauser, 2009). Three aspects
of the community are considered: demographics, public services, and a
general community evaluation. For demographics, we include the sex
ratio (the percentage female) and the number of big surnames (a surname
shared by large numbers of people in the village). The latter is included
since the possibility exists that some unobserved genetic factors systemat -
ically affect people’s cognitive functions, and the number of big surnames
may indicate relatives in the community with whom the respondent may
interact so it may serve as a measure of social connectedness.
Community-level variables generated from the CHARLS commu-
nity survey are used to capture public services. These community-level
variables are derived from a separate survey of community leaders that
describes current facilities and histories of the community. For example,
the number of libraries and the distance from the community/village
office to the most commonly used library are both used. Similarly, num-
ber and distance are asked for kindergartens, primary schools, middle
schools, senior high schools, theatres, nursing homes, bus lines, and train
station. We include a variable measuring percentage of people with one or
more telephones and percentage with one or more mobile phones in the
community. Accessibility and availability of public services is a reflection
of social and economic capital in a society that is effective in improving
people’s health (Yip et al., 2007). For evaluation of community, we have
three measures: (1) log public expenditure during the past year, (2) com-
munity log per capita expenditures, and (3) interviewer rating of whether
community medical service is poor. Community log PCE is an average of
individual respondents’ log PCEs in each community.
Summary Statistics
Table 12-1 presents summary statistics for participation in social activi-
ties by sex, region, province, and age group. Fewer than half of respon-
dents engaged in any kind of social activities, with 44% of women and 47%
of men taking part. Interacting with friends is the most popular activity,
with a participation rate of 34.5%, followed by Mahjong, chess, cards, or a
club (17.1%), and only 8.6% participating in “other social activities.”6 Men
are more active in playing Mahjong, chess, cards, or going to a club, doing
voluntary or charity work, or attending an educational or training course,
while women are more inclined to interact with friends.
Regional disparity is more pronounced than gender differences. Liv -
6 There is some overlap among the various social activities.
OCR for page 289
289
YUQING HU, XIAOYAN LEI, JAMES P. SMITH, and YAOHUI ZHAO
ing in an urban area or in Zhejiang province has a great advantage in
participation in social activities: 53.5% of urban people took part in social
activities, compared with only 39.4% for those in rural areas. In Zhejiang,
55.6% of residents are involved in social activities, compared to only
33.8% of Gansu residents. Urban residents have a 7.4% higher participa -
tion rate in interacting with friends, and a 10.4% higher rate in playing
Mahjong, chess, or cards. This rural-urban difference may be due partly
to the urban elderly having more leisure time than rural elderly who are
busy trying to make a living. One factor that likely impacts participation
is accessibility and availability of activating facilities in the village/com -
munity, as discussed below.
The last four columns in Table 12-1 list social activity participation
across four age groups. People become less engaged as they age, which
may be partly due to declining health with age.7 However, there are two
exceptions: participation rates for interacting with friends drop sharply
between ages 50–60 (from 40 to 31.1%). But then it bounces back up in
the 60–70 age group (32.3%), indicating that while retirement may reduce
interactions with work-related social networks, after a period of adjust -
ment, people are able to socialize with friends outside of their former jobs.
Taking part in a community-related organization demonstrates a similar
pattern. This may be explained by the possibility that retirement detaches
people from occupational positions in community-related organizations,
but they once again become involved in such organizations as partici -
pants rather than staff.
Table 12-3 summarizes the distribution of test scores on the MI and
EM of respondents, where large differences exist across sex, region,
and age. The mean score for MI is 8.62 and for EM is 2.99. The description
illustrates that women are on average much more cognitively impaired.
Rural Gansu elderly are also more disadvantaged in both aspects of cogni-
tive abilities. Comparing people of different age groups, the older group is
associated with lower cognitive ability, which is consistent with cognition
declining with aging.
Table 12-4 provides a summary of demographic and SES attributes
across sex, region, province, and age groups. Our analytical sample has a
mean age of 59.4, is about half female, 44.4% urban residents, and 54.6%
Zhejiang residents. Age, sex, and location of residence do not signifi-
cantly differ among these groups. Respondents have an average of 2.74
children and 3.21 siblings. Female, rural, and Gansu elderly have more
7 These age patterns may reflect cohort effects as well, and cohort effects are plausible given
the large changes in China over time. With only cross-sectional data, we cannot distinguish
between age and cohort effects. The use of 10-year age groups centered on ages 50 and 60
reflects the fact that the official retirement age for urban women is 50 and for urban men is 60.
OCR for page 296
296 AGING IN ASIA
think that the female population share represents the level of economic
and social conditions in local areas. As shown in Table 12-5, although the
share of female population is normal overall (50.1%), the maximum share
reaches 53.3% and the lowest 48.6%. The unbalanced gender ratio is due
to two reasons. First, in areas of longer life expectancy, women outlive
men to create a higher ratio of females in the population. Second, in less
developed areas where sons are valued more than daughters, sex-selec -
tive abortion produced highly skewed gender ratio among the younger
population (Ebenstein, 2010). In both cases, a higher ratios of females is
associated with better life conditions, which may be positively correlated
with community cognition levels. Other community-level variables may
not be able to fully capture these regional disparities. With prefecture
fixed effects (not shown), the significance of percentage female disap -
pears, supporting the hypotheses here.9
Table 12-7 illustrates a parallel OLS association estimation in which
EM is regressed on variables representing social activities and the same
set of other control variables included in the MI model.10 Both any social
activity participation collectively and interacting with friends or playing
card games are strongly statistically associated with EM. In addition, in
the second model that includes the single measure of participation in any
social activity it is strongly related to our measure of episodic memory.
Looking next at the control variables, once again memory increases
with age at a decreasing rate. Memory level reaches its peak at
47 ( = (0.134 * 100)/(2 * 0.142)) years of age, after which it starts declin -
ing. Unlike the disparity between men and women for mental intactness,
there is generally no sex difference for EM and no statistically significant
relationship with the number of relatives. All regional measures—urban
and province—are also not significant, suggesting that the other covari -
ates (mostly likely education and income) pick up the regional disparities
displayed in the descriptive tables. Once again, education and log PCE are
strongly related to our cognitive measure—in this case, episodic memory.
From the bottom panel, most community-level variables are statisti -
cally insignificant. Similar to the prior table, distance between community
center and most-commonly used middle school is negatively correlated
with episodic memory. Episodic memory is positively associated with
the number of kindergartens in the community. These results indicate
that availability and access to education resources may play important
roles in improving people’s mental health. Taking account of unobserved
A prefecture is an administrative district that is between province and county
9
or county-level city.
10 As the two dependent variables have different missing values, observations of
these two regressions are slightly different.
OCR for page 297
297
YUQING HU, XIAOYAN LEI, JAMES P. SMITH, and YAOHUI ZHAO
TABLE 12-7 OLS Analysis of Effects of SA on EM
Dependent Variable: Episodic Memory
(0–10)
Independent Variables (1) (2)
Social Activities (no participation omitted)
Interacting with friends 0.222** (0.088)
Playing Mahjong, chess, or cards 0.459*** (0.110)
Other social activities 0.186 (0.149)
Any of the above activities 0.404*** (0.084)
Demographics
Age (45+) 0.133*** (0.049) 0.134*** (0.049)
Age squared/100 –0.140*** (0.040) –0.142*** (0.040)
Female 0.036 (0.091) 0.004 (0.090)
Urban 0.157 (0.118) 0.155 (0.118)
Zhejiang –0.191 (0.147) –0.196 (0.147)
No. of children –0.028 (0.035) –0.026 (0.035)
No. of siblings 0.000 (0.022) –0.002 (0.022)
SES
Education (illiterate omitted)
Sishu/home school and below 0.610*** (0.118) 0.632*** (0.118)
Elementary school 0.866*** (0.123) 0.877*** (0.123)
Middle school 1.541*** (0.141) 1.562*** (0.140)
High school and above 1.508*** (0.165) 1.564*** (0.163)
Log PCE 0.160*** (0.053) 0.169*** (0.052)
Community
% of female 0.012 (0.007) 0.013* (0.007)
# of big surnames 0.085** (0.039) 0.089** (0.039)
# of kindergartens 0.067* (0.039) 0.068* (0.039)
Distance from kindergarten –0.006 (0.007) –0.006 (0.007)
# of primary schools –0.008 (0.124) –0.006 (0.124)
Distance from primary school 0.051 (0.060) 0.047 (0.060)
# of middle schools –0.069 (0.142) –0.035 (0.142)
Distance from middle school –0.041** (0.018) –0.040** (0.018)
# of senior high schools –0.002 (0.260) –0.021 (0.261)
Distance from senior high school 0.002 (0.005) 0.002 (0.005)
# of post offices 0.224 (0.155) 0.203 (0.155)
Distance from post office –0.006 (0.013) –0.007 (0.013)
# of libraries 0.146* (0.081) 0.153* (0.081)
Distance from the library 0.005 (0.006) 0.005 (0.006)
# of theatres –0.126 (0.190) –0.112 (0.190)
Distance from the theatre –0.005 (0.005) –0.006 (0.005)
# of nursing homes –0.228 (0.176) –0.248 (0.177)
Distance from nursing home 0.001 (0.004) 0.002 (0.004)
# of bus lines 0.004 (0.013) 0.004 (0.013)
Distance from bus line 0.006 (0.010) 0.006 (0.010)
Distance from train station 0.000 (0.001) 0.001 (0.001)
% of people having telephones 0.003 (0.002) 0.003 (0.002)
% of people having cellphones –0.001 (0.002) –0.001 (0.002)
Log public expenditure during the past year –0.019** (0.009) –0.018** (0.009)
Log per capita income 0.136** (0.061) 0.142** (0.061)
Whether the medical care is poor 0.041 (0.356) 0.041 (0.356)
Observations 1,901 1,901
R-squared 0.279 0.276
NOTE: Standard errors in parentheses.* denotes p < 0.1; ** p < 0.05; *** p < 0.01.
SOURCE: Data from CHARLS 2008 Pilot.
OCR for page 298
298 AGING IN ASIA
prefecture level variations in culture, climate, and geographic factors that
may affect cognition, we also estimated prefecture fixed effects models.
The results (not shown) are remarkably similar to those contained in
Tables 12-6 and 12-7. The significance and magnitudes of coefficients on
SA do not change much.
Summarizing our results, OLS association regressions with multiple
SA measures provide evidence of a positive correlation between cogni-
tive functioning and participation in social activities. Despite the salient
association, closer scrutiny is needed because OLS regressions may be
biased due to the endogeneity of people’s participation decisions, an issue
presented in the next section.
THE ASSOCIATION OF COMMUNITY FACILITIES
AND COGNITIVE FUNCTION
OLS estimates of the association between social activities and cogni-
tive functions do not have a causal interpretation as unobserved heteroge-
neity may introduce a correlation between social activities and cognitive
function. Sources of this unobserved heterogeneity may include aspects of
the environment, people’s preferences, and personalities. While the natural
environment is controllable to some extent by including community-level
variables in the models, the other sources of individual heterogeneity are
not easy to control. Although we include a rich set of individual and spa -
tial variables in our model, they cannot be perfectly controlled for. Another
possible source of bias is reverse causality—taking part in social activities
requires some minimum cognitive skills or people with cognitive impair-
ment deliberately seek social activities as therapy for slowing down the
decline.
One option to get around this problem is to find instrumental vari -
ables. We have a set of instruments in mind, i.e., the availability of com-
munity facilities that accommodate social gatherings. For example, in
1995, the Chinese government started the Universal Exercise Plan (1995–
2010) encouraging Chinese people to participate in physical activities.
In the past decade, especially during pre-Olympic years, enforcement
of the plan was strengthened. Accompanying this was a larger govern-
mental investment in community recreational facilities/organizations and
advocacy for participating in exercises. This development is different
across community/village because of variation in the capacity for build -
ing facilities and the strength of local leaders. Because there is no evi -
dence that this difference is directly related to cognition function, these
variables can serve as candidates for IVs for social activities. However,
probably because the number of communities in our sample is relatively
small, these variables do not pass weak instrument tests. Thus, instead of
OCR for page 299
299
YUQING HU, XIAOYAN LEI, JAMES P. SMITH, and YAOHUI ZHAO
directly tackling the causal relationship, we opt to estimate OLS reduced
form equations to test whether these variables are associated with cogni -
tive function.
The CHARLS community survey contains information on whether
a village/community has certain facilities/organizations shown on a
list. There are 15 items on the list: (1) a basketball court, (2) a swim -
ming pool, (3) outside exercising facilities, (4) outdoor sports facilities,
(5) rooms for card games and chess games, (6) rooms for playing ping-
pong, (7) calligraphy and painting associations, (8) other entertainment
facilities, (9) dancing teams or other exercise organizations, (10) organi-
zations for helping the elderly and the frail, (11) employment services,
(12) healthcare organizations, (13) activity centers for elderly, (14) elderly
associations, and (15) nursing homes.
The first variable of community facility, “outdoor facilities,” takes
value 1 if the community has outdoor recreational facilities (1–4 on the
list). Having outdoor facilities provide people with place(s) to gather.
The second variable, “whether there are activity centers for the elderly”
(item 13) is necessary for people to play Mahjong, chess, or cards, or par-
ticipate in a club. The third variable is the dummy indicating whether the
community has organizations that help the elderly and the handicapped
(item 10). We also create a variable indicating whether the community
has any of the three types of facilities. The items in the survey ques -
tion not utilized include employment service, healthcare organizations,
elderly associations, and nursing homes. Though important facilities and
organizations, these items are omitted due to their possible direct impact
on people’s cognitive health (for instance, employment services teach
people job skills and knowledge that improves cognition, and healthcare
organizations enhance mental intactness and cure cognitive diseases such
as dementia). As the purpose of examining the facilities is to explore the
effect of social activities on cognition, we only keep those that are likely
to affect cognition through affecting social activities. The detailed distri -
butions of these facility variables by residency and province are reported
in Appendix Table 12-A1.
Table 12-3 depicts the relationship between facility variables and their
corresponding SAs, showing that the facility variables are highly associ-
ated with SA participation among the elderly, consistent with the mecha-
nisms discussed above. Tables 12-8 and 12-9 give the OLS reduced form
estimates of the associations of community facilities and the two types
of cognitive functions, MI and EM, respectively. Similarly as before, we
employ two models in each table. Model 1 includes the three types of
facilities separately, while Model 2 only the aggregate facility measure
“any of the above facilities.” From Table 12-8, the facility measures do
not significantly affect mental intactness, which is consistent with the
OCR for page 300
300 AGING IN ASIA
results in the OLS association estimation in Table 12-6, where almost all
social activity measures lack significance on the correlation with mental
intactness.
Table 12-9 replaces MI as the dependent variable with EM. As seen
from the first model, having activity centers for the elderly is significantly
(at 1% level) associated with better episodic memory, while the other two
types of facilities are not. These results are also consistent with the OLS
association estimates in Table 12-6 where we see significant (at 1% level)
correlation between playing Mahjong, chess, or cards. The likely expla-
nation is that having activity centers for the elderly allowing for more
involvement in playing Mahjong, chess, or cards, which further promotes
episodic memory. Accordingly, from the second model, we see significant
correlation between any of the facilities and episodic memory, which is
most likely driven by the effect of having activity centers for the elderly.
In sum, we interpret our results as providing some support for the
hypothesis that taking part in social activities could help slow down
the cognitive decline of the elderly. The results are consistent with find-
ings from Glei et al. (2005) that employed a similar definition and was
based on data from Taiwan, where the tradition of living arrangement
is most similar. This indicates that social participation plays an impor-
tant role even in a society where family members are more likely to live
together, regardless of level of economic development. Our results also
imply that providing more activity facilities in the community could be
helpful to the cognitive health of the elderly as facilities involve more
people in social activities. We are cautious in our interpretation since
we do not directly estimate the causal effects especially in light of the
relatively small sizes in the CHARLS pilot, which was fielded in only
two provinces. Most sample sizes in the HRS network of surveys exceed
10,000 households, and China is a very heterogeneous country, which
makes larger samples even more important. The availability of the full
CHARLS survey, especially its panel, in a few years with much higher
samples will help provide additional tests of our hypothesis.
CONCLUSIONS
With data from the 2008 CHARLS pilot, this chapter explores the
relation between SA and cognitive outcomes for Chinese elderly aged 45
and older, where cognition is composed of two dimensions—MI and EM.
There are several key limitations to our analysis. Most important, we are
only able to examine two Chinese provinces with moderate sample sizes
and must rely for now on cross-sectional analysis. A larger sample size
spanning the entire country would not only address the issue of repre-
sentativeness, but also provide greater variation across communities—the
OCR for page 301
301
YUQING HU, XIAOYAN LEI, JAMES P. SMITH, and YAOHUI ZHAO
TABLE 12-8 OLS Reduced Form Analysis of Community Facilities on MI
Dependent Variable: Mental Intactness
(0–11)
Independent Variables (1) (2)
Community Facilities
Outdoor facilities 0.042 (0.166)
Activity centers for the elderly 0.028 (0.125)
Organizations for helping the elderly and the –0.069 (0.128)
handicapped
Any of the above –0.046 (0.128)
Demographics
Age (45+) 0.128*** (0.048) 0.129*** (0.048)
Age squared/100 –0.135*** (0.039) –0.135*** (0.039)
Female –0.499*** (0.087) –0.499*** (0.087)
Urban 0.121 (0.117) 0.117 (0.115)
Zhejiang 0.736*** (0.166) 0.760*** (0.146)
No. of children 0.056 (0.035) 0.056 (0.035)
No. of siblings –0.011 (0.021) –0.011 (0.021)
SES
Education (illiterate omitted)
Sishu/home school and below 1.294*** (0.115) 1.296*** (0.115)
Elementary school 1.504*** (0.120) 1.506*** (0.120)
Middle school 1.827*** (0.137) 1.826*** (0.137)
High school and above 2.313*** (0.158) 2.311*** (0.158)
Log PCE 0.135*** (0.052) 0.139*** (0.051)
Community
% of female 0.022*** (0.007) 0.022*** (0.007)
# of big surnames –0.039 (0.039) –0.038 (0.038)
# of kindergartens 0.075* (0.041) 0.069* (0.038)
Distance from kindergarten 0.004 (0.007) 0.004 (0.007)
# of primary schools 0.053 (0.123) 0.056 (0.121)
Distance from primary school 0.101* (0.060) 0.113* (0.060)
# of middle schools 0.016 (0.141) 0.001 (0.137)
Distance from middle school –0.062*** (0.018) –0.062*** (0.018)
# of senior high schools –0.189 (0.277) –0.133 (0.250)
Distance from senior high school –0.001 (0.005) –0.001 (0.005)
# of post offices 0.012 (0.160) 0.037 (0.155)
Distance from post office 0.003 (0.013) 0.002 (0.013)
# of libraries 0.092 (0.079) 0.089 (0.079)
Distance from the library –0.012** (0.006) –0.012** (0.005)
# of theatres –0.165 (0.187) –0.173 (0.186)
Distance from the theatre –0.004 (0.004) –0.005 (0.004)
# of nursing homes 0.032 (0.176) 0.049 (0.173)
Distance from nursing home 0.007* (0.004) 0.007* (0.003)
# of bus lines 0.025* (0.013) 0.022* (0.012)
Distance from bus line 0.007 (0.010) 0.008 (0.010)
Distance from train station 0.000 (0.001) 0.000 (0.001)
% of people having telephones 0.001 (0.002) 0.001 (0.002)
% of people having cellphones 0.000 (0.002) 0.000 (0.002)
Log public expenditure during the past year 0.000 (0.009) –0.000 (0.009)
Log per capita income –0.012 (0.067) –0.010 (0.061)
Whether the medical care is poor –0.552 (0.353) –0.519 (0.346)
Observations 1,873 1,873
R-squared 0.396 0.396
NOTE: Standard errors in parentheses.* denotes p < 0.1; ** p < 0.05; *** p < 0.01.
SOURCE: Data from CHARLS 2008 Pilot.
OCR for page 302
302 AGING IN ASIA
TABLE 12-9 OLS Reduced Form Analysis of Effects of Community
Facilities on EM
Dependent Variable: Episodic Memory
(0–10)
Independent Variables (1) (2)
Community Facilities
Outdoor facilities 0.266 (0.171)
Activity centers for the elderly 0.408*** (0.128)
Organizations for helping the elderly and the –0.022 (0.132)
handicapped
Any of the above 0.540*** (0.131)
Demographics
Age (45+) 0.129*** (0.049) 0.130*** (0.049)
Age squared/100 –0.139*** (0.040) –0.141*** (0.040)
Female –0.010 (0.090) –0.014 (0.090)
Urban 0.142 (0.120) 0.133 (0.119)
Zhejiang –0.239 (0.168) –0.231 (0.149)
No. of children –0.019 (0.036) –0.018 (0.036)
No. of siblings 0.001 (0.022) 0.002 (0.022)
SES
Education (illiterate omitted)
Sishu/home school and below 0.615*** (0.119) 0.614*** (0.119)
Elementary school 0.887*** (0.123) 0.878*** (0.123)
Middle school 1.570*** (0.141) 1.544*** (0.141)
High school and above 1.635*** (0.163) 1.631*** (0.162)
Log PCE 0.183*** (0.053) 0.184*** (0.052)
Community
% of female 0.010 (0.007) 0.009 (0.007)
# of big surnames 0.068* (0.040) 0.076* (0.039)
# of kindergartens 0.091** (0.042) 0.046 (0.039)
Distance from kindergarten –0.005 (0.008) –0.005 (0.008)
# of primary schools 0.037 (0.126) 0.028 (0.124)
Distance from primary school 0.017 (0.061) –0.021 (0.062)
# of middle schools 0.045 (0.146) –0.034 (0.142)
Distance from middle school –0.037** (0.019) –0.033* (0.019)
# of senior high schools –0.359 (0.289) –0.131 (0.262)
Distance from senior high school 0.002 (0.005) 0.006 (0.005)
# of post offices 0.051 (0.163) 0.088 (0.158)
Distance from post office –0.002 (0.013) –0.002 (0.013)
# of libraries 0.171** (0.081) 0.191** (0.082)
Distance from the library 0.005 (0.006) 0.006 (0.006)
# of theatres –0.111 (0.192) –0.178 (0.190)
Distance from the theatre –0.006 (0.005) –0.008* (0.005)
# of nursing homes –0.390** (0.182) –0.336* (0.178)
Distance from nursing home 0.001 (0.004) 0.002 (0.004)
# of bus line 0.003 (0.013) 0.006 (0.013)
Distance from bus line 0.001 (0.010) 0.003 (0.010)
Distance from train station 0.001 (0.001) 0.001 (0.001)
% of people having telephones 0.002 (0.002) 0.002 (0.002)
% of people having cellphones 0.000 (0.002) 0.001 (0.002)
Log public expenditure during the past year –0.017* (0.009) –0.016* (0.009)
Log per capita income 0.099 (0.068) 0.101 (0.062)
Whether the medical care is poor –0.071 (0.364) –0.003 (0.357)
Observations 1,901 1,901
R–squared 0.272 0.274
NOTE: Standard errors in parentheses.* denotes p < 0.1; ** p < 0.05; *** p < 0.01.
SOURCE: Data from CHARLS 2008 Pilot.
OCR for page 303
303
YUQING HU, XIAOYAN LEI, JAMES P. SMITH, and YAOHUI ZHAO
analytical center of our analysis. Panel samples will aid in identification
of causal pathways.
Our major findings are roughly half of Chinese aged 45 and older
take part in social activities. Second, participation in social activities is
highly dependent on respondents’ socioeconomic status and community
environment. Third, OLS association results show that playing Mahjong,
chess, or card games and interacting with friends are significantly related
to episodic memory, both individually and taken as a whole (any of the
three activities); individually, they are not related to mental intactness,
while taken as a whole they are. Fourth, having an activity center in the
community is significantly related to higher episodic memory but has
no relation to mental intactness. These results point to a possible causal
relationship between social activities and cognitive function, especially in
strengthening short-term memory. Our research suggests that an effective
way to maintain cognitive abilities at advanced ages may be to improve
community facilities, such as by providing Mahjong rooms or other enter-
tainment facilities.
APPENDIX TABLE 12-A1 Distribution of Community Facilities
Region Province
Instrumental Variables All Urban Rural Zhejiang Gansu
Outdoor facilities 0.101 0.161 0.053 0.080 0.126
(.301) (.368) (.223) (.271) (.332)
Activity centers for the elderly 0.496 0.698 0.335 0.753 0.187
(.500) (.459) (.472) (.431) (.390)
Organizations for helping the 0.307 0.528 0.130 0.335 0.272
elderly and the handicapped (.461) (.500) (.337) (.472) (.445)
For any SA: Any of the facilities 0.591 0.824 0.405 0.814 0.322
above (.492) (.381) (.491) (.389) (.468)
Number of facilities 0.904 1.387 0.518 1.168 0.585
(.901) (.901) (.690) (.770) (.943)
Observations 2,293 1,017 1,276 1,253 1,040
NOTE: Standard errors in parentheses.* denotes p < 0.1; ** p < 0.05; *** p < 0.01.
SOURCE: Data from CHARLS 2008 Pilot.
OCR for page 304
304 AGING IN ASIA
REFERENCES
Allaire, J.C., and M. Marsiske. (1999). Everyday cognition: Age and intellectual ability
correlates. Psychology and Aging 14(4):627-644. Available: http://www.ncbi.nlm.nih.
gov/pmc/articles/PMC2904910/.
China Center for Economic Research, Peking University. (2008). China Health and Retirement
Longitudinal Survey. Available: http://charls.ccer.edu.cn/charls/data.asp.
Ebenstein, A. (2010). The “missing girls” of China and the unintended consequences of the
one child policy. Journal of Human Resources 45:87-115.
Engelhardt, H., I. Buber, V. Skirbekk, and A. Prskawetz. (2010). Social involvement,
behavioural risks and cognitive functioning among the aged. Ageing and Society 30.
doi: 10.1017/S0144686X09990626. Available: http://www.share-austria.at/fileadmin/
user_upload/articles/2010_Cognitivefunctioning.pdf.
Glei, D.A., D.A. Landau, N. Goldman, Y-L. Chuang, G. Rodríguez, and M. Weinstein.
(2005). Participating in social activities helps preserve cognitive function. Interna-
tional Journal of Epidemiology 34(4):864-871. Available: http://www.ncbi.nlm.nih.gov/
pubmed/15764689.
Hauser, R.M. (2009). Causes and consequences of cognitive functioning across the life course.
Educational Researcher 39(2):95-109.
Hsu, H.C. (2007). Does social participation by the elderly reduce mortality and cognitive
impairment? Aging & Mental Health 11(6):699-707.
Lei, X., Y. Hu, J.J. McArdle, J.P. Smith, and Y. Zhao. (2011). Gender Differences in Cognition
Among Older Adults in China. Working Paper #WR-881. Available: http://www.rand.
org/content/dam/rand/pubs/working_papers/2011/RAND_WR881.pdf.
Leibovici, D., K. Ritchie, B. Ledesert, and D. Touchon. (1996). Does education level determine
the course of cognitive decline? Age and Ageing 25:392-397.
Levy, R. (1994). Aging-associated cognitive decline. International Psychogeriatrics 663-668.
Available: http://journals.cambridge.org/action/displayAbstract?fromPage=online&
aid=272198.
Löckenhoff, C.E., and L.L. Carstensen. (2004). Socio-emotional selectivity theory, aging, and
health: The increasingly delicate balance between regulating emotions and making
tough choices. Journal of Personality 72(6):1,395-1,424. Available: http://www.ncbi.nlm.
nih.gov/pubmed/15509287.
McArdle, J.J., J.P. Smith, and R. Willis. (2011). Cognition and economic outcomes in the
Health and Retirement Survey. Pp. 209-236 in Explorations in the Economics of Aging,
D. Wise (Ed.). Chicago: University of Chicago Press.
Schaie, K.W. (1994). The course of adult intellectual development. American Psychologist
49:304-313.
Smith, J.P., Y. Shen, J. Strauss, Z. Yang, and Y. Zhao. (2012). The effects of childhood health
on adult health and SES in China. Economic Development and Cultural Change.
Strauss, J., and D. Thomas, D. (2008). Health over the life course. In Handbook of Development
Economics, Volume 4, T.P. Schultz and J. Strauss (Eds.). Amsterdam: North Holland
Press.
Tilvis, R.S., M.H. Kähönen-Väre, J. Jolkkonen, J. Valvanne, K.H. Pitkala, and T.E. Strandberg.
(2004). Predictors of cognitive decline and mortality of aged people over a 10-year
period. The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences 59(3):
268-274. Available: http://www.ncbi.nlm.nih.gov/pubmed/15031312.
Verhaegen, P., and T.A. Salthouse. (1997). Meta-analyses of age-cognition relations in adult -
hood. Estimates of linear and nonlinear age effects and structural models. Psychological
Bulletin 122(3):231-249.
OCR for page 305
305
YUQING HU, XIAOYAN LEI, JAMES P. SMITH, and YAOHUI ZHAO
Wang, H.X., A. Karp, B. Winblad, and L. Fratiglioni. (2002). Decreased risk of dementia: A
longitudinal study from the Kungsholmen Project. American Journal of Epidemiology
155(12):1,081-1,087.
Yip, W., S.V. Subramanian, A.D. Mitchell, T.S.L. Dominic, J. Wang, and I. Kawachi. (2007).
Does social capital enhance health and well-being? Evidence from rural China. Social
Science & Medicine 64:35-49.
Zhao, Y., J. Strauss, A. Park, Y. Shen, and Y. Sun. (2009). Chinese Health and Retirement
Longitudinal Study, Pilot, User’s Guide. Peking: Peking University, National School of
Development.
Zunzunegui, M-V., B.E. Alvarado, T. Del Ser, and A. Otero. (2003). Social networks, social
integration, and social engagement determine cognitive decline in community-dwelling
Spanish older adults. The Journals of Gerontology. Series B, Psychological Sciences and Social
Sciences 58(2):S93-S100. Available: http://www.ncbi.nlm.nih.gov/pubmed/12646598.
OCR for page 306
