Skip to main content
NCBI home page
The Journals of Gerontology Series B: Psychological Sciences and Social Sciences logoLink to The Journals of Gerontology Series B: Psychological Sciences and Social Sciences
. 2023 Jun 9;78(10):1756–1764. doi: 10.1093/geronb/gbad088

Education, Social Engagement, and Cognitive Function: A Cross-Lagged Panel Analysis

Chenguang Du 1, Yasuo Miyazaki 2, XinQi Dong 3, Mengting Li 4,✉
Editor: Jessica Kelley
PMCID: PMC10561888  PMID: 37294899

Abstract

Objectives

Although education and social engagement are considered cognitive reserves, the pathway of both reserves on cognitive function has been rarely studied. This study aimed to examine the underlying mechanism between education, social engagement, and cognitive function.

Methods

This study used 2-wave data (2010 and 2014) from Health and Retirement Study in the United States (N = 3,201). Education was measured by years of schooling. Social engagement was evaluated by 20 items including volunteering, physical activities, social activities, and cognitive activities. Cognitive function was assessed by a modified Telephone Interview for Cognitive Status. A cross-lagged panel model was fitted to test the mediating mechanism between education, social engagement, and cognitive function.

Results

Controlling for covariates, higher education in early life was associated with better cognitive function in old age (b = 0.211, 95% confidence interval [CI] = [0.163, 0.259], p < .01). Late-life social engagement partially mediated the association between education and cognitive function (indirect effect = 0.021, 95% CI = [0.010, 0.033], p < .01). The indirect path between education and social engagement via cognition also existed (b = 0.009, 95% CI = [0.005, 0.012], p < .001).

Discussion

Education in earlier life stage may exert a lifelong effect on cognitive function as well as an indirect effect via enhancing late-life cognitive reserve such as social engagement. The cross-lagged effect of social engagement on cognitive function is significant and vice versa. Future research may explore other cognitive reserves over the life course and its underlying mechanism to achieve healthy cognitive aging.

Keywords: Health and Retirement Study, Life course perspective, Mediating effect

As a key element of successful aging, maintaining high cognitive capacity was of essential importance for independent living and quality of life in late adulthood (Rowe & Kahn, 1997). Individual differences exist in susceptibility to age-related brain changes or Alzheimer’s disease-related pathology (Song et al., 2022; Stern et al., 2020). The capacity to tolerate brain changes or pathology before it reaches a critical threshold for clinical symptoms was defined as reserve, and two types of reserve were proposed by Stern and colleagues (2020): brain reserve and cognitive reserve. Brain reserve refers to actual differences in the brain itself that may increase tolerance of pathology, while cognitive reserve focuses on individual differences in how tasks are processed that can allow some individuals to cope better than others with brain changes (Stern et al., 2020). Reserve is not a fixed entity but can change across the life span depending on exposures and behaviors (Stern, 2012).

Past decades have witnessed a great effort to identify lifetime risk and protective factors to achieve healthy cognitive aging (Livingston et al., 2017). Empirical studies indicated that older adults with higher educational attainment in early life (Gatz et al., 2001; van Hooren et al., 2007) and higher levels of social engagement in later life (Lövdén et al., 2005; Podewils et al., 2005) tended to show slower cognitive decline or lower risks of Alzheimer’s disease. The pathways model suggests that the impact of early-life conditions on health in later life is mediated by later exposures. In line with the pathways model, social engagement could be potentially important pathways linking education to late-life cognition. Only few studies investigated the mediating role of social engagement in the relationship between education and cognitive function but have been restricted by nonnationwide samples (Zhang et al., 2019), cross-sectional data (Solis-Urra et al., 2020), and conceptualization and measurement of lifestyle factors (Liu & Lachman, 2019; Solis-Urra et al., 2020).

On the one hand, education could have a direct and indirect effect on cognitive function via social engagement. On the other hand, higher levels of education could be associated with higher levels of social engagement via better cognitive function. This reciprocal relationship between late-life social engagement and cognitive function has not yet been tested in prior studies (Liu & Lachman, 2019; Zhang et al., 2019). To address these gaps, this study aims to test the underlying mechanism between education, social engagement, and cognitive function with a cross-lagged panel model using the two-wave Health and Retirement Study (HRS).

Passive and Active Models of Reserve

The passive (threshold) models focus on the “hardware” of neural function (e.g., brain size or head circumference), while the active models concentrate on the “software” of brain processing (e.g., education and occupational attainment; Stern et al., 2020). In line with the passive model, brain reserve is defined as the amount of damage that can be sustained before reaching a threshold for clinical expression and greater reserve is protective simply by creating distance from the cutoff threshold of functioning for dementia. In terms of the active model, cognitive reserve refers to the ability to recruit alternative brain networks to compensate for the effects of pathology (Stern, 2002). The active and passive models are not mutually exclusive. Individuals with higher education have greater brain weight and larger neurons (Katzman et al., 1988), resulting in reduced risks of Alzheimer’s disease.

Cognitive Reserve in Early Life: Education

Education is one of the widely used proxy measures for the active models of cognitive reserve. Individuals with higher education may process tasks in a more efficient manner (Stern, 2002). The active models were supported by a large body of empirical studies. Studies consistently showed a reduced risk of developing dementia among older adults with higher level of education (Livingston et al., 2017; Song et al., 2022). Similarly, subsequent research found that individuals with higher education tend to have a better performance across a broad range of cognitive tasks. For example, a study in Netherlands indicated that older adults with middle and high education level significantly outperformed cognitive tasks such as executive function, verbal fluency, verbal memory, and cognitive speed tasks than their counterparts with low level of education (van Hooren et al., 2007). In a longitudinal study, Zahodne et al. (2011) found that participants’ educational attainment was positively associated with working memory, verbal fluency, and episodic memory.

Some evidence has shown a paradoxical manifestation of cognitive reserve. Higher cognitive reserve (e.g., higher education) may delay the onset of cognitive impairment, but the transition between mild cognitive impairment and dementia was relatively steep. Individuals with higher cognitive reserve can tolerate more pathology, so the point at which cognitive function begins to be affected will be later than in those with lower cognitive reserve. However, there is a common point in all people where the pathology is so severe that function cannot be maintained. Individuals with higher cognitive reserve will begin their cognitive decline when pathology is more advanced and thus have less time until they reach the point where pathology overwhelms function. This results in a more rapid rate of decline once it begins (Stern, 2012). Results showed that higher education was related to a shorter period between the onset of accelerated decline and the dementia diagnosis (Hall et al., 2007; van Loenhoud et al., 2019).

Modifiable Lifestyle Factor in Later Life: Social Engagement

The cognitive reserve theory suggested that an engaged lifestyle (e.g., reading, writing, or regular exercise) facilitates the practice of various cognitive skills, which plays a protective role against cognitive decline (Fratiglioni et al., 2004). Active older adults with more frequent contacts and integration could have higher self-esteem, resulting in lower stress and better cognitive health (Fratiglioni et al., 2004). Experimental studies in rats have shown that environmentally enriched conditions have the potential to prevent or reduce their cognitive deficits (Pham et al., 2002). Observational cross-sectional studies showed that older adults who were more engaged in social activities tend to have better performance on cognitive function (Fu et al., 2018; Sakamoto et al., 2017). Longitudinal studies found that increased social engagement in later life was associated with slower cognitive decline (Bourassa et al., 2017; Lövdén et al., 2005) as well as decreased risk of dementia (Saczynski et al., 2006; Zhou et al., 2018).

Life Course Epidemiological Models of Pathways From Childhood Education to Late-Life Health

A life course perspective provides a useful lens with which to understand the impact of earlier life events on subsequent events (Elder, 1994). Epidemiologists have used life course perspectives to conceptualize the relationship between early-life conditions and health outcomes in later life (Kuh et al., 2003). Life course epidemiology has received increasing attention in the field of cognitive aging (Ben-Shlomo et al., 2016; Kuh et al., 2003). It indicated that the connection between different phases of life could be viewed as a pathways model, wherein the circumstances in early adulthood may affect social integration and cognitive health later in life (Greenfield et al., 2021; Kuh et al., 2003). In line with this perspective, early-life education may become physiologically embedded in ways that directly influence late-life cognitive function or change the trajectory of exposures experienced later in life (e.g., active lifestyle) and thus exert indirect influence (Lachman et al., 2010; Thrane, 2006).

Varied pathways to late adulthood through education are likely to produce developmental variations. The differences in early-life education bear upon subsequent options and social engagement that differentiate the experience of aging in later life. Education may cultivate health literacy and resources necessary for continued engagement in intellectually demanding activities (e.g., read books, use internet) or health-promoting behaviors (e.g., physical activity) in old age.

Recent empirical studies demonstrated the potential mediating role of lifestyle factors in the relationship between education and cognition in older adulthood (Liu & Lachman, 2019; Solis-Urra et al., 2020; Zhang et al., 2019). For example, a cross-sectional study by Solis-Urra et al. (2020) found that leisure-time physical activity could partially mediate the association between education and cognitive impairment among older Chileans. Using a national database of midlife in the United States, Liu and Lachman (2019) found that both physical and cognitive activities could partially mediate the longitudinal association between educational attainment and cognitive ability. Similar results also existed among older Chinese Americans (Zhang et al., 2019), where both participants’ social and cognitive activities could partially mediate the association between education and global cognitive function as well as its various domains.

On the one hand, education could influence cognitive function through social engagement. On the other hand, an alternative pathway may exist, and education may affect social engagement via cognitive function. Evidence has shown that cognitive frailty was significantly associated with low social engagement (Wada et al., 2022). Small et al. (2012) found that activity participation influenced subsequent change in cognitive performance and vice versa, which suggested a reciprocal relationship between late-life social engagement and cognition. Although a few studies tested the former pathway (Liu & Lachman, 2019; Zhang et al., 2019), the latter pathway that education may affect social engagement through cognitive function has not yet been addressed in previous studies.

The Present Study

The reviewed literature reveals that the underlying relationship between education, social engagement, and cognitive function has not been fully explored. To fill the research gaps, this study examines the mechanism between education, social engagement, and cognitive function with two-wave data from HRS, which is a large-scale nationwide data set. The cross-lagged panel model was adopted to address the potential reciprocal relationship between social engagement and cognitive function. In line with the active model of reserve, we hypothesized that: (1) individuals with more educational attainment are associated with better cognitive function. Based on the life course framework and prior empirical research, we also hypothesized that: (2) late-life social engagement could partially mediate the association between education and late-life cognitive function. In addition, we hypothesized the alternative pathway: (3) late-life cognitive function could partially mediate the association between education and late-life social engagement. Figure 1 illustrates the conceptual framework.

Figure 1.

Figure 1.

Open in a new tab

Conceptual framework of cross-lagged associations among education, social engagement, and cognitive function.

Method

Sample and Data

The current study used two-wave longitudinal data (2010 and 2014) from HRS, which is a longitudinal panel study of adults aged 50 and older in the United States. The core survey collected rich information including participants’ sociodemographic, financial, physical, and psychological status. In addition, the HRS used Leave-Behind Questionnaire (LBQ) to collect participants’ psychosocial and lifestyle information biennially (Taylor & Nguyen, 2020).

Social participation items were added to the LBQ in 2008, 2010, 2012, 2014, and 2016. The 2010 and 2014 LBQs used subsample A, while the 2008, 2012, and 2016 LBQs used subsample B. The 2008 LBQ used 18 items to measure social engagement. The 2010 and 2014 surveys used the same 20 items to assess social engagement and the response categories were changed to 7-point Likert scale. The 2016 LBQ used 21 items. Given the variations in the measurement and sample, we selected 2010 and 2014 data, which used the same sample and measurements across waves. Our data set was obtained by merging the RAND (the RAND HRS data file is a cleaned and easy-to-use longitudinal data, which contains most used HRS variables) longitudinal data (V1) with LBQ in 2010 and 2014.

The eligibility criteria included: (1) respondents who were 65 or older in 2010 because some of the cognitive tests (e.g., object naming) were only assessed in participants aged 65 or older; (2) respondents who have been interviewed in both 2010 and 2014. In other words, respondents who died between 2010 and 2014 (or were lost to follow-up) were excluded. The final longitudinal sample included 3,201 older adults.

Measures

Global cognitive function

Participants’ global cognitive function was measured by a modified Telephone Interview for Cognitive Status (Griffin et al., 2020). The measurement was administered for all participants in each wave, which assessed their immediate and delayed word recall, backward counting, serial subtraction, and object naming. The score of all the assessed items was summed to create a composite score with the range from 0 to 35. A high composite score indicated a better global cognitive function. Its psychometric property has been validated in prior research (Herzog & Wallace, 1997).

Social engagement

Social engagement was assessed by 20 items with a wide range of activities including volunteering, physical activities (e.g., sports or exercise), social activities (e.g., community or interest group meeting), and cognitive activities (e.g., read books, or use internet; Luster et al., 2022). The responses were rated by a 7-point Likert scale (1 = Daily, 2 = Several times a week, 3 = Once a week, 4 = Several times a month, 5 = At least once a month, 6 = Not in the last month, 7 = Never/Not relevant). Each item score was reverse-coded so that a higher score indicated a higher level of social engagement. A composite score of social engagement was calculated by averaging across all 20 items. The Cronbach alphas for social engagement in 2010 and 2014 are 0.73 and 0.75, respectively.

Education and covariates

The independent variable was participants’ educational attainment measured by years of education, which was treated as an exogenous variable. Covariates included participants’ age in 2010, gender (0 = male; 1 = female), race/ethnicity (White, Black, and others), total wealth measured as total wealth components less all debt, depressive symptoms, and chronic conditions. The depressive symptoms were measured by the eight-item Center for Epidemiological Studies—Depression scale. A sum score of these eight items was obtained to indicate the participants’ level of depression. The Cronbach alpha for depressive symptoms in 2010 and 2014 was 0.77 and 0.78, respectively. Chronic conditions were the total number of doctor-diagnosed chronic health conditions, including high blood, stroke, cancer or a malignant tumor, chronic lung disease, arthritis, cardiovascular disease, diabetes, emotional or psychiatric problems, and sleep disorders.

Data Analysis

Descriptive analyses for all study variables were conducted. To explore the mediating mechanism among education, social engagement, and global cognitive function, we adopted a two-wave cross-lagged panel model (Cole & Maxwell, 2003). Compared with the mediating test in cross-sectional data, the two-wave cross-lagged model excels in taking into account the temporal sequence of measured variables as well as autoregression of measured variables at low cost of time and money, which is more suitable for estimating for the potential causal effect (Zhao et al., 2021). In Figure 1, educational attainment was an exogenous variable measured before other variables as a time-constant variable. The path coefficient “c” represents the direct effect of early-life education on cognitive function in 2014 (Hypothesis 1). The path coefficient “a” represents the direct effect of early-life education on social engagement in 2010. The path “b” indicates the direct effect of social engagement in 2010 on cognitive function in 2014, controlling for covariates and previous level of cognitive function in 2010. Therefore, the indirect effect of education on cognitive function in 2014 is estimated as the product term between path a and b (Little et al., 2007). Hypothesis 2 could be tested by the significance test for a × b. Another possible path that will be examined simultaneously is from education to social engagement in 2014, which is mediated by cognition in 2010 (Hypothesis 3). Likewise, the corresponding direct effects among those variables were indicated by aʹ, bʹ, and cʹ.

Goodness-of-fit indexes including chi-square test, comparative fit index (CFI), Standardized Root Mean Square Residual (SRMR), and root mean square error of approximation (RMSEA), which were used as criteria for evaluating the model fit. A minimum CFI value of 0.90 and a value less than 0.05 for RMSEA and 0.08 for SRMR were considered a good model fit for this study (Xia & Yang, 2019). A bootstrapping method (10,000 samples) was applied to obtain standard error (SE) and confidence intervals (CIs) for both direct and indirect effects, and all the analysis will be conducted by Mplus version 7.4 (Muthén & Muthén, 1998).

The psychosocial leave-behind respondent weight in 2010, which adjusted for selection into the face-to-face sample and nonresponse to the psychosocial questionnaire, was applied to all structural equation models in the current study.

Results

Sample Characteristics

Table 1 illustrates descriptive information for all the study variables. Participants’ average age in 2010 and 2014 was 74.18 years (standard deviation [SD] = 6.29) and 77.85 years (SD = 6.29), respectively. More than half of the participants (58%) were female (n = 1,871), and 86% of them were White (n = 2,750). The average educational level was 12.75 years (SD = 2.89). In terms of racial composition, most participants were White (86%), with 11% Black and 3% others. Regarding marital status, the percentage of the married decreased from 62% in 2010 to 56% in 2014. In contrast, the average number of chronic conditions increased from 2.30 in 2010 to 2.67 in 2014, indicating that the average health conditions of respondents became worse. Similarly, participants’ average depressive symptoms increased from 1.13 to 1.24. The average score of cognition in 2010 was 22.28 (SD = 4.39). The median score of cognition in 2010 was 23.00 (interquartile range [IQR] = 5.00). The average score of cognition in 2014 was 21.20 (SD = 5.04). The median score of cognition in 2014 was 22.00 (IQR = 7.00). According to poverty thresholds provided by United States Census Bureau (2023; i.e., $13,194 for 2010 and $14,326 for 2014), there were around 11.9% of older adults living in poverty in 2010, and this ratio was increased to 12.7% in 2014.

Table 1.

Descriptive Statistics for Study Variables Across Waves

Variables 2010 2014
Gender, n (%)
 Male 1,330 (42%)
 Female 1,871 (58%)
Race, n (%)
 White 2,750 (86%)
 Black 344 (11%)
 Others 107 (3%)
Marital status, n (%)
 Married 1,991 (62%) 1,788 (56%)
 Other status 1,210 (38%) 1,410 (44%)
Age, mean (SD) 74.18 (6.29) 77.85 (6.29)
Education, mean (SD) 12.75 (2.89)
Total wealth, mean (SD) 517,760.68 (821,462.51) 556,673.89 (984,310.97)
Chronic conditions, mean (SD) 2.30 (1.35) 2.67 (1.41)
Depressive symptoms, mean (SD) 1.13 (1.69) 1.24 (1.78)
Social engagement, mean (SD) 3.39 (0.80) 3.22 (0.81)
Cognitive function, mean (SD) 22.28 (4.39) 21.20 (5.04)
Open in a new tab

Notes: Cognitive function in 2010: median (IQR) = 23.00 (5.00). Cognitive function in 2014: median (IQR) = 22.00 (7.00); IQR = interquartile range; SD = standard deviation.

Path Models for Education, Social Engagement, and Global Cognition

Figure 2 shows the estimates for the main paths among interested variables. The model fit indices indicated that the hypothesized model fitted the data well: CFI = 0.962, RMSEA = 0.054, SRMR = 0.060. Chi-square test was statistically significant (χ2 = 437.779, df = 43, p < .05), which is common as its value tends to increase along with the sample size (Barrett, 2007).

Figure 2.

Figure 2.

Open in a new tab

Path coefficients for the cross-lagged mediation model among older adults over 65.

The solid lines in Figure 2 represent statistically significant paths among interested variables with unstandardized regression coefficient and SE attached along each path. Education was positively associated with social engagement in 2010 (b = 0.073, SE = 0.005, p < .001), controlling for baseline age, gender, race, marital status, depression, total wealth, and chronic conditions. This result suggested that older adults who had higher educational attainment in early adulthood tended to have higher level of social engagement in older adulthood. Furthermore, social engagement in 2010 was positively associated with cognitive function in 2014 (b = 0.295, SE = 0.099, p = .003), controlling for cognitive function in 2010 and covariates listed in the former step. This lagged effect suggested that individuals with higher level of social engagement would have better cognitive function in later life compared with those who have lower level of social engagement. Finally, the direct effect of education in early adulthood on late-life cognitive function was also statistically significant (b = 0.214, SE = 0.026, p < .001), implying that individuals with a higher educational level tended to have better cognitive function in later life even after controlling for social engagement in 2010 and other covariates.

In terms of the alternative pathway, the results indicated that education was positively associated with cognitive function in 2010 (b = 0.583, SE = 0.027, p < .001), which was positively related to social engagement in 2014 (b = 0.015, SE = 0.003, p < .001). In addition, the association between education and social engagement was also statistically significant (b = 0.020, SE = 0.006, p = .002).

Testing the Significance of Indirect and Total Effects

Table 2 shows the significance test for the indirect and total effect of mediational models shown in Figure 1. For the hypothesized model, the indirect effect of education on cognitive function in 2014 was statistically significant (b = 0.021, 95% CI = [0.007, 0.036], p = .004). The total effect from education to cognitive function in 2014 was statistically significant (b = 0.236, 95% CI = [0.186, 0.286], p < .001). These results indicated that the social engagement could partially mediate the association between education and cognitive function in current study.

Table 2.

Estimate for the Direct and Indirect Effect Among Older Adults Over 65

Regression coefficient
Coefficients 95% CI
Direct effect
 Education → cognitive function 0.214*** [0.163, 0.266]
 Education → social engagement 0.020** [0.007, 0.032]
Indirect effect
 Education → cognitive function 0.021** [0.007, 0.036]
 Education → social engagement 0.009*** [0.005, 0.013]
Total effect
 Education → cognitive function 0.236*** [0.186, 0.286]
 Education → social engagement 0.028** [0.016, 0.041]
Open in a new tab

Notes: CI = confidence interval. CI values represent 95% bias-corrected, bootstrapped CIs.

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

Interestingly, the indirect effect of education on social engagement via cognitive function (b = 0.009, 95% CI = [0.005, 0.013], p < .001) and its total effect (b = 0.028, 95% [0.016, 0.041], p < .01) were also statistically significant. However, the effect size was much smaller than the mediational path between education and cognitive function via social engagement regarding both indirect and total effects (see Table 2).

Additional Analyses

As cognitive function scores tend to remain stable in the younger old and decrease markedly after 75 years of age, which is around the mean age of the sample, it is worth testing whether social engagement continues to be beneficial after 75 years of age and to assess if early-life education continues to have a direct effect at those advanced ages.

We fitted the same hypothetical model in a subsample aged 75 and older (n = 1,363). The model fit indices showed that the model fitted the data well: CFI = 0.975, RMSEA = 0.055, SRMR = 0.055, χ2 = 179.463 (df = 43, p < .05). In addition, both the direct (b = 0.141, 95% CI = [0.060, 0.222], p < .001) and indirect effects (b = 0.028, 95% CI = [0.005, 0.052], p < .01) of education on cognitive function in 2014 were statistically significant (Supplementary Appendixes 1 and 2).

As more than 80% of our sample were older White adults, we fitted the same hypothetical model in a subsample with only older White adults (n = 2,750). The results indicated that the model fitted the data well: CFI = 0.973, RMSEA = 0.055, SRMR = 0.054. Chi-square test was statistically significant (χ2 = 259.395, df = 29, p < .05). Both the direct (b = 0.206, 95% CI = [0.149, 0.263], p < .001) and indirect (b = 0.024, 95% CI = [0.007, 0.041], p < .01) effects of education on cognition in 2014 were significant (Supplementary Appendixes 3 and 4).

Longitudinal study of cognitive function among older adults is often subject to attrition where respondents dropped out from the study (due to death or poor health) in between the period from 2010 and 2014. Attrition analysis, which used a logistic regression, indicated that compared with those who completed survey in both waves, the dropout sample in the current study tends to be older, Black, and male, with lower educational attainment, more chronic health conditions, and higher depressive symptoms (Supplementary Appendix 5).

Discussion

This study examined social engagement as the mechanism linking education to cognitive function among older adults in the United States using a nationwide data set. We found a positive relationship between educational attainment and cognitive function, indicating that individuals who received higher education tended to have better cognitive function in later adulthood. Moreover, this association was partially mediated by social engagement, as evidenced by the significant indirect path from education to cognitive function through social engagement. Participants who received higher education in early life reported higher levels of social engagement in later adulthood, which further helped maintain a better cognitive function. In addition, this study demonstrated the alternative mechanism that education could affect social engagement (in 2014) via prior cognition (in 2010). Our findings added evidence regarding the complex relationship between education, social engagement, and cognitive function.

In line with prior empirical studies (Cagney & Lauderdale, 2002; Wilson et al., 2009), our findings showed that higher education in early life was associated with better cognitive function in older adulthood. Therefore, our first hypothesis was supported. Education is one of the first and most commonly used proxies in studies on cognitive reserve (Nucci et al., 2012). An individual’s education is influenced by broader socioeconomic circumstances and exposures, which may affect late-life health (Lyu & Burr, 2016; Pudrovska & Anikputa, 2014; Walsemann et al., 2016). In addition, our results demonstrated the additive effects of education and social engagement with respect to late-life cognition, which partially supported the accumulation model (Pudrovska & Anikputa, 2014; Walsemann et al., 2016). Future research could further test the interactive effects of education and social engagement.

The mediating role of social engagement in the association between education and cognitive function was identified and thus our second hypothesis was supported. The Lancet Commission on Dementia Prevention, Intervention, and Care identified lifetime modifiable factors to manage and prevent dementia and it is suggested that both an increase in childhood education and an increase in late-life social engagement could contribute to prevention or delay of dementia (Livingston et al., 2017). In line with prior studies (Liu & Lachman, 2019; Solis-Urra et al., 2020; Zhang et al., 2019), our finding supported the pathways model. Cognitive reserves are potentially intercorrelated (Steffener & Stern, 2012). One cognitive reserve may enhance the other. Compared with the poorly educated, well-educated individuals are more likely to engage or develop healthier lifestyles such as physical exercise or cognitively stimulating activities, which may protect individuals’ overall health.

The coexisting mediational path from education to social engagement via cognitive function may imply a complex dynamic relationship among the three variables. Social engagement, as a sociobehavioral proxy of cognitive reserve, has been well-studied in the literature (Stern et al., 2020), while the impact of cognitive function on social engagement received less attention. Cognitive impairment prevents older adults from engaging in an active lifestyle (Barrett et al., 2011). This study strengthens the idea that social engagement and cognitive function mutually influence each other. Education is a predictor for the reciprocal social engagement–cognition relationship. Higher education could benefit both cognitive health and social well-being in later life.

Our study has several limitations to be considered for further research. First, the two-wave data with a 4-year gap do not allow us to examine the effect of education on cognitive change over a longer period of time. In addition, two-wave data reduced the opportunities from using more advanced analytical methods such as random-intercept cross-lagged model (RI-CLM; Hamaker et al., 2015). By taking advantage of the multilevel structure where observation points are nested within persons, the RI-CLM will account for not only temporal stability, but also for trait-like stability, which help depict the cross-lagged relationship within the individual level. Therefore, further analysis may consider using three or more waves of data and advanced methods to improve the estimation. Second, although HRS is designed to be representative for the noninstitutionalized U.S. population of older adults over age 50 (McArdle et al., 2007; Sonnega et al., 2014), our study used a subsample of HRS (from the left-behind survey), which may influence its representativeness. White, highly educated, and wealthy individuals may have been overrepresented in our analytic sample, as we can see that 86% of our sample in 2010 was classified as White. This overrepresentation could have occurred for the following reasons. In our analytic sample, older adults who died between 2010 and 2014 (or were lost to follow-up) were excluded. Prior HRS research indicated that African Americans, unhealthy, and less-educated respondents are more likely to die over the study period (Michaud et al., 2011). Conversely, White, highly educated, and wealthy group of participants could have higher retention rates in the sample because this group could have higher survival rates (Attanasio & Hoynes, 2000). Given these reasons, our findings might not be generalizable to older adults from ethnic minorities, with low education and income. In fact, attrition analysis using a logistic regression indicated that the dropout sample in the current study tended to be older, Black, and male, with lower education, more chronic health condition, and higher depressive symptoms, compared with those who completed the survey in both waves. Therefore, we infer that our results, which ignored these dropout older adults, may have created potential bias, probably downward bias. This is because the loss of less-educated people would reduce the size of the positive effects of education on both cognition and social engagement.

Our study has some research and practical implications. This study incorporated a life course perspective and tested a sequence of linked protective factors that lead to late-life cognitive outcomes, while addressing reverse causality. The findings indicated that education in an earlier life stage may exert a long-term beneficial effect on cognitive function. Meanwhile, the significant indirect effect of the modifiable protective factor, social engagement, was identified by the pathways model suggesting that cognitive reserves emerging throughout the life course interconnectedly affected late-life cognition. Socioeconomic disparities in childhood might lead to health disparities in later life. Interventions on improving social engagement could reduce negative impacts of low education and promote successful aging. Our findings showed that the indirect effect of education on cognitive function via social engagement only accounted for 9% of the total effect. This may imply that other potential mediators in the association between education and cognitive function might exist. This study only tested the active model of reserve. Future research could examine the passive model of reserve as well as the relationship between active and passive models of reserve. Future studies could also explore the impact of mild, moderate, and severe cognitive impairment to see whether the protection role holds in different stages of cognitive decline.

Conclusion

This study was among the first to use cross-lagged panel model to examine the direct and indirect effects of education on cognition through social engagement as well as the alternative mediating path. Our findings indicated that older adults with higher education tended to maintain better cognitive function partially through increased social engagement. The results implied further work may explore other modifiable resources that could explain the association between education and cognition among older adults. Given the reciprocal relationship between social engagement and cognitive function, it is also important for practitioners to develop multidisciplinary intervention programs to increase the level of social engagement and protect cognitive function for older adults.

Supplementary Material

gbad088_suppl_Supplementary_Appendix
Click here for additional data file. (1MB, pdf)

Acknowledgments

The authors appreciated Nader Mehri, PhD, from UNC School of Social Medicine for the advice on the manuscript revision.

Contributor Information

Chenguang Du, School of Medicine, University of North Carolina in Chapel Hill, Chapel Hill, North Carolina, USA.

Yasuo Miyazaki, School of Education, Virginia Polytechnic Institute and State University, Blacksburg, Virginia, USA.

XinQi Dong, Institute for Health, Health Care Policy and Aging Research, Rutgers, The State University of New Jersey, New Brunswick, New Jersey, USA.

Mengting Li, Department of Social Security, School of Labor and Human Resources, Renmin University of China, Beijing, China.

Funding

M. Li was supported by Alzheimer’s Association (AARG-NTF-20-684568). The Health and Retirement Study is sponsored by the National Institute on Aging (grant number NIA U01AG009740) and is conducted by the University of Michigan.

Conflict of Interest

None.

Author Contributions

M. Li and Y. Miyazaki planned the study and supervised the data analysis. C. Du performed all statistical analyses. C. Du and M. Li wrote the paper. Y. Miyazaki and X. Dong contributed to revising the paper.

References

  1. Attanasio, O. P., & Hoynes, H. W. (2000). Differential mortality and wealth accumulation. The Journal of Human Resources, 35(1), 1–29. doi: 10.2307/146354 [DOI] [Google Scholar]
  2. Barrett, A., Burke, H., Cronin, H., Hickey, A., Kamiya, Y., Kenny, R. A., Layte, R., Maty, S., McGee, H., & Morgan, K. (2011). Fifty plus in Ireland 2011: First results from the Irish Longitudinal Study on Ageing (TILDA). Dublin: Trinity College Dublin, 10, 2011–2000. doi: 10.25419/rcsi.10770389.v2 [DOI] [Google Scholar]
  3. Barrett, P. (2007). Structural equation modelling: Adjudging model fit. Personality and Individual Differences, 42(5), 815–824. doi: 10.1016/j.paid.2006.09.018 [DOI] [Google Scholar]
  4. Ben-Shlomo, Y., Cooper, R., & Kuh, D. (2016). The last two decades of life course epidemiology, and its relevance for research on ageing. International Journal of Epidemiology, 45(4), 973–988. doi: 10.1093/ije/dyw096 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bourassa, K. J., Memel, M., Woolverton, C., & Sbarra, D. A. (2017). Social participation predicts cognitive functioning in aging adults over time: Comparisons with physical health, depression, and physical activity. Aging & Mental Health, 21(2), 133–146. doi: 10.1080/13607863.2015.1081152 [DOI] [PubMed] [Google Scholar]
  6. Cagney, K. A., & Lauderdale, D. S. (2002). Education, wealth, and cognitive function in later life. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 57(2), 163–172. doi: 10.1093/geronb/57.2.P163 [DOI] [PubMed] [Google Scholar]
  7. Cole, D. A., & Maxwell, S. E. (2003). Testing mediational models with longitudinal data: Questions and tips in the use of structural equation modeling. Journal of Abnormal Psychology, 112(4), 558–577. doi: 10.1037/0021-843X.112.4.558 [DOI] [PubMed] [Google Scholar]
  8. Elder, G. H. (1994). Time, human agency, and social change: Perspectives on the life course. Social Psychology Quarterly, 57(1), 4–15. doi: 10.2307/2786971 [DOI] [Google Scholar]
  9. Fratiglioni, L., Paillard-Borg, S., & Winblad, B. (2004). An active and socially integrated lifestyle in late life might protect against dementia. Lancet Neurology, 3(6), 343–353. doi: 10.1016/s1474-4422(04)00767-7 [DOI] [PubMed] [Google Scholar]
  10. Fu, C., Li, Z., & Mao, Z. (2018). Association between social activities and cognitive function among the elderly in China: A cross-sectional study. International Journal of Environmental Research and Public Health, 15(2), 231. doi: 10.3390/ijerph15020231 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Gatz, M., Svedberg, P., Pedersen, N. L., Mortimer, J. A., Berg, S., & Johansson, B. (2001). Education and the risk of Alzheimer’s disease: Findings from the study of dementia in Swedish twins. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 56(5), 292–300. doi: 10.1093/geronb/56.5.p292 [DOI] [PubMed] [Google Scholar]
  12. Greenfield, E. A., Moorman, S., & Rieger, A. (2021). Life course pathways from childhood socioeconomic status to later-life cognition: evidence from the Wisconsin Longitudinal Study. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 76(6), 1206–1217. doi: 10.1093/geronb/gbaa062 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Griffin, S. C., Mezuk, B., Williams, A. B., Perrin, P. B., & Rybarczyk, B. D. (2020). Isolation, not loneliness or cynical hostility, predicts cognitive decline in older Americans. Journal of Aging and Health, 32(1), 52–60. doi: 10.1177/0898264318800587 [DOI] [PubMed] [Google Scholar]
  14. Hall, C. B., Derby, C., LeValley, A., Katz, M. J., Verghese, J., & Lipton, R. B. (2007). Education delays accelerated decline on a memory test in persons who develop dementia. Neurology, 69(17), 1657–1664. doi: 10.1212/01.wnl.0000278163.82636.30 [DOI] [PubMed] [Google Scholar]
  15. Hamaker, E. L., Kuiper, R. M., & Grasman, R. P. (2015). A critique of the cross-lagged panel model. Psychological Methods, 20(1), 102–116. doi: 10.1037/a0038889 [DOI] [PubMed] [Google Scholar]
  16. Herzog, A. R., & Wallace, R. B. (1997). Measures of cognitive functioning in the AHEAD Study. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 52(Special_Issue), 37–48. doi: 10.1093/geronb/52B.Special_Issue.37 [DOI] [PubMed] [Google Scholar]
  17. Katzman, R., Terry, R., DeTeresa, R., Brown, T., Davies, P., Fuld, P., Renbing, X., & Peck, A. (1988). Clinical, pathological, and neurochemical changes in dementia: A subgroup with preserved mental status and numerous neocortical plaques. Annals of Neurology, 23(2), 138–144. doi: 10.1002/ana.410230206 [DOI] [PubMed] [Google Scholar]
  18. Kuh, D., Ben-Shlomo, Y., Lynch, J., Hallqvist, J., & Power, C. (2003). Life course epidemiology. Journal of Epidemiology and Community Health, 57(10), 778–783. doi: 10.1136/jech.57.10.778 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Lachman, M. E., Agrigoroaei, S., Murphy, C., & Tun, P. A. (2010). Frequent cognitive activity compensates for education differences in episodic memory. The American Journal of Geriatric Psychiatry, 18(1), 4–10. doi: 10.1097/JGP.0b013e3181ab8b62 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Little, T. D., Preacher, K. J., Selig, J. P., & Card, N. A. (2007). New developments in latent variable panel analyses of longitudinal data. International Journal of Behavioral Development, 31(4), 357–365. doi: 10.1177/0165025407077757 [DOI] [Google Scholar]
  21. Liu, Y., & Lachman, M. E. (2019). Education and cognition in middle age and later life: The mediating role of physical and cognitive activity. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 75(7), e93–e104. doi: 10.1093/geronb/gbz020 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Livingston, G., Sommerlad, A., Orgeta, V., Costafreda, S. G., Huntley, J., Ames, D., Ballard, C., Banerjee, S., Burns, A., Cohen-Mansfield, J., Cooper, C., Fox, N., Gitlin, L. N., Howard, R., Kales, H. C., Larson, E. B., Ritchie, K., Rockwood, K., Sampson, E. L., … Mukadam, N. (2017). Dementia prevention, intervention, and care. The Lancet (British Edition), 390(10113), 2673–2734. doi: 10.1016/S0140-6736(17)31363-6 [DOI] [PubMed] [Google Scholar]
  23. Lövdén, M., Ghisletta, P., & Lindenberger, U. (2005). Social participation attenuates decline in perceptual speed in old and very old age. Psychology and Aging, 20(3), 423–434. doi: 10.1037/0882-7974.20.3.423 [DOI] [PubMed] [Google Scholar]
  24. Luster, J. E., Ratz, D., & Wei, M. Y. (2022). Multimorbidity and social participation is moderated by purpose in life and life satisfaction. Journal of Applied Gerontology, 41(2), 560–570. doi: 10.1177/07334648211027691 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lyu, J., & Burr, J. A. (2016). Socioeconomic status across the life course and cognitive function among older adults: An examination of the latency, pathways, and accumulation hypotheses. Journal of Aging and Health, 28(1), 40–67. doi: 10.1177/0898264315585504 [DOI] [PubMed] [Google Scholar]
  26. McArdle, J. J., Fisher, G. G., & Kadlec, K. M. (2007). Latent variable analyses of age trends of cognition in the Health and Retirement Study, 1992–2004. Psychology and Aging, 22(3), 525–545. doi: 10.1037/0882-7974.22.3.525 [DOI] [PubMed] [Google Scholar]
  27. Michaud, P.-C., Kapteyn, A., Smith, J. P., & van Soest, A. (2011). Temporary and permanent unit non-response in follow-up interviews of the Health and Retirement Study. Longitudinal and Life Course Studies, 2, 145–169. doi: 10.14301/llcs.v2i2.114 [DOI] [Google Scholar]
  28. Muthén, L. K., & Muthén, B. O. (1998). 1998–2011. Mplus user’s guide. Muthén & Muthén. [Google Scholar]
  29. Nucci, M., Mapelli, D., & Mondini, S. (2012). Cognitive Reserve Index questionnaire (CRIq): A new instrument for measuring cognitive reserve. Aging Clinical and Experimental Research, 24, 218–226. doi: 10.3275/7800 [DOI] [PubMed] [Google Scholar]
  30. Pham, T. M., Winblad, B., Granholm, A.-C., & Mohammed, A. H. (2002). Environmental influences on brain neurotrophins in rats. Pharmacology, Biochemistry and Behavior, 73(1), 167–175. doi: 10.1016/s0091-3057(02)00783-9 [DOI] [PubMed] [Google Scholar]
  31. Podewils, L. J., Guallar, E., Kuller, L. H., Fried, L. P., Lopez, O. L., Carlson, M., & Lyketsos, C. G. (2005). Physical activity, APOE genotype, and dementia risk: Findings from the Cardiovascular Health Cognition Study. American Journal of Epidemiology, 161(7), 639–651. doi: 10.1093/aje/kwi092 [DOI] [PubMed] [Google Scholar]
  32. Pudrovska, T., & Anikputa, B. (2014). Early-life socioeconomic status and mortality in later life: An integration of four life-course mechanisms. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 69(3), 451–460. doi: 10.1093/geronb/gbt122 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Rowe, J. W., & Kahn, R. L. (1997). Successful aging. Gerontologist, 37(4), 433–440. doi: 10.1093/geront/37.4.433 [DOI] [PubMed] [Google Scholar]
  34. Saczynski, J. S., Pfeifer, L. A., Masaki, K., Korf, E. S. C., Laurin, D., White, L., & Launer, L. J. (2006). The effect of social engagement on incident dementia: The Honolulu–Asia Aging Study. American Journal of Epidemiology, 163(5), 433–440. doi: 10.1093/aje/kwj061 [DOI] [PubMed] [Google Scholar]
  35. Sakamoto, A., Ukawa, S., Okada, E., Sasaki, S., Zhao, W., Kishi, T., Kondo, K., & Tamakoshi, A. (2017). The association between social participation and cognitive function in community-dwelling older populations: Japan Gerontological Evaluation Study at Taisetsu community Hokkaido. International Journal of Geriatric Psychiatry, 32(10), 1131–1140. doi: 10.1002/gps.4576. [DOI] [PubMed] [Google Scholar]
  36. Small, B. J., Dixon, R. A., McArdle, J. J., & Grimm, K. J. (2012). Do changes in lifestyle engagement moderate cognitive decline in normal aging? Evidence from the Victoria Longitudinal Study. Neuropsychology, 26(2), 144–155. doi: 10.1037/a0026579 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Solis-Urra, P., Plaza-Diaz, J., Álvarez-Mercado, A. I., Rodríguez-Rodríguez, F., Cristi-Montero, C., Zavala-Crichton, J. P., Olivares-Arancibia, J., Sanchez-Martinez, J., & Abadía-Molina, F. (2020). The mediation effect of self-report physical activity patterns in the relationship between educational level and cognitive impairment in elderly: A cross-sectional analysis of Chilean Health National Survey 2016–2017. International Journal of Environmental Research and Public Health, 17(8), 2619. doi: 10.3390/ijerph17082619 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Song, S., Stern, Y., & Gu, Y. (2022). Modifiable lifestyle factors and cognitive reserve: A systematic review of current evidence. Ageing Research Reviews, 74, 101551. doi: 10.1016/j.arr.2021.101551 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Sonnega, A., Faul, J. D., Ofstedal, M. B., Langa, K. M., Phillips, J. W., & Weir, D. R. (2014). Cohort profile: The Health and Retirement Study (HRS). International Journal of Epidemiology, 43(2), 576–585. doi: 10.1093/ije/dyu067 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Steffener, J., & Stern, Y. (2012). Exploring the neural basis of cognitive reserve in aging. Biochimica et Biophysica Acta (BBA)—Molecular Basis of Disease, 1822(3), 467–473. doi: 10.1016/j.bbadis.2011.09.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Stern, Y. (2002). What is cognitive reserve? Theory and research application of the reserve concept. Journal of the International Neuropsychological Society, 8(3), 448–460. doi: 10.1017/s1355617702813248 [DOI] [PubMed] [Google Scholar]
  42. Stern, Y. (2012). Cognitive reserve in ageing and Alzheimer’s disease. The Lancet Neurology, 11(11), 1006–1012. doi: 10.1016/S1474-4422(12)70191-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Stern, Y., Arenaza-Urquijo, E. M., Bartrés-Faz, D., Belleville, S., Cantilon, M., Chetelat, G., Ewers, M., Franzmeier, N., Kempermann, G., Kremen, W. S., Okonkwo, O., Scarmeas, N., Soldan, A., Udeh-Momoh, C., Valenzuela, M., Vemuri, P., Vuoksimaa, E., & The Reserve, Resilience and Protective Factors PIA Empirical Definitions and Conceptual Frameworks Workgroup. (2020). Whitepaper: Defining and investigating cognitive reserve, brain reserve, and brain maintenance. Alzheimer’s & Dementia, 16(9), 1305–1311. doi: 10.1016/j.jalz.2018.07.219 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Taylor, H. O., & Nguyen, A. W. (2020). Depressive symptoms and loneliness among Black and White older adults: The moderating effects of race. Innovation in Aging, 4(5) , 1–13. doi: 10.1093/geroni/igaa048 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Thrane, C. (2006). Explaining educational-related inequalities in health: Mediation and moderator models. Social Science and Medicine, 62(2), 467–478. doi: 10.1016/j.socscimed.2005.06.010 [DOI] [PubMed] [Google Scholar]
  46. United States Census Bureau. (2023). Poverty thresholds. https://www.census.gov/data/tables/time-series/demo/income-poverty/historical-poverty-thresholds.html
  47. van Hooren, S. A. H., Valentijn, A. M., Bosma, H., Ponds, R. W. H. M., van Boxtel, M. P. J., & Jolles, J. (2007). Cognitive functioning in healthy older adults aged 64–81: A cohort study into the effects of age, sex, and education. Aging Neuropsychology and Cognition, 14(1), 40–54. doi: 10.1080/138255890969483 [DOI] [PubMed] [Google Scholar]
  48. van Loenhoud, A. C., van der Flier, W. M., Wink, A. M., Dicks, E., Groot, C., Twisk, J., Barkhof, F., Scheltens, P., Ossenkoppele, R., & Alzheimer's Disease Neuroimaging Initiative. (2019). Cognitive reserve and clinical progression in Alzheimer disease: A paradoxical relationship. Neurology, 93(4), e334–e346. doi: 10.1212/WNL.0000000000007821 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Wada, A., Makizako, H., Nakai, Y., Tomioka, K., Taniguchi, Y., Sato, N., Kiuchi, Y., Kiyama, R., Kubozono, T., Takenaka, T., & Ohishi, M. (2022). Association between cognitive frailty and higher-level competence among community-dwelling older adults. Archives of Gerontology and Geriatrics, 99, 104589. doi: 10.1016/j.archger.2021.104589 [DOI] [PubMed] [Google Scholar]
  50. Walsemann, K. M., Goosby, B. J., & Farr, D. (2016). Life course SES and cardiovascular risk: Heterogeneity across race/ethnicity and gender. Social Science & Medicine, 152, 147–155. doi: 10.1016/j.socscimed.2016.01.038 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Wilson, R. S., Hebert, L. E., Scherr, P. A., Barnes, L. L., De Leon, C. F. M., & Evans, D. A. (2009). Educational attainment and cognitive decline in old age. Neurology, 72(5), 460–465. doi: 10.1212/01.wnl.0000341782.71418.6c [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Xia, Y., & Yang, Y. (2019). RMSEA, CFI, and TLI in structural equation modeling with ordered categorical data: The story they tell depends on the estimation methods. Behavior Research Methods, 51(1), 409–428. doi: 10.3758/s13428-018-1055-2 [DOI] [PubMed] [Google Scholar]
  53. Zahodne, L. B., Glymour, M. M., Sparks, C., Bontempo, D., Dixon, R. A., MacDonald, S. W. S., & Manly, J. J. (2011). Education does not slow cognitive decline with aging: 12-year evidence from the Victoria Longitudinal Study. Journal of the International Neuropsychological Society, 17(6), 1039–1046. doi: 10.1017/S1355617711001044 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Zhang, W., Tang, F., Chen, Y., Silverstein, M., Liu, S., & Dong, X. (2019). Education, activity engagement, and cognitive function in US Chinese older adults. Journal of the American Geriatrics Society, 67(S3), S525–S531. doi: 10.1111/jgs.15560 [DOI] [PubMed] [Google Scholar]
  55. Zhao, X., Jin, L., & Sun, S. B. (2021). The bidirectional association between physical and cognitive function among Chinese older adults: A mediation analysis. International Journal of Aging & Human Development, 92(2), 240–263. doi: 10.1177/0091415020940214 [DOI] [PubMed] [Google Scholar]
  56. Zhou, Z., Wang, P., & Fang, Y. (2018). Social engagement and its change are associated with dementia risk among Chinese older adults: A longitudinal study. Scientific Reports, 8(1), 1551. doi: 10.1038/s41598-017-17879-w [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

gbad088_suppl_Supplementary_Appendix
Click here for additional data file. (1MB, pdf)

Articles from The Journals of Gerontology Series B: Psychological Sciences and Social Sciences are provided here courtesy of Oxford University Press

RESOURCES