Spousal Education and Cognitive Functioning in Later Life

Abstract

Objectives

Numerous studies have documented the relationship between education and cognitive functioning at the individual level. Yet few studies have examined whether a spouse’s education spills over to influence the other spouse’s cognitive functioning. This study, therefore, investigates the association between spousal education and cognitive functioning, the pathways that may account for this association, and gender differences in this association.

Method

Growth curve models were analyzed by using longitudinal couple data from the Health and Retirement Study (N = 5,846 individuals).

Results

More years of spousal education are associated with higher level of cognitive functioning at age 65 (γ₀₀₀ = 0.0532, 95% confidence interval [CI] = 0.0163−0.0901) and slower decline in cognitive functioning in later adulthood (γ₁₀₀ = 0.0054, 95% CI = 0.0026−0.0082). The positive association between spousal education and the level of cognitive functioning at age 65 is fully explained by economic resources. The association of spousal education with the rate of change in cognitive functioning decreases but remains significant after controlling for economic resources and health behaviors (γ₁₀₀ = 0.0043, 95% CI = 0.0014−0.0072). The association between spousal education and cognitive functioning is similar for men and women.

Discussion

Findings suggest that more years of spousal education may slow decline in cognitive functioning for men and women in later life.

Cognitive functioning—which generally declines with advancing age—is a key determinant of independence and quality of life in later adulthood (Medalia & Erlich, 2017). Severe decline in cognitive functioning increases risk of comorbidity and mortality, demands intensive caregiving from relatives, and escalates health care expenditures. Given the continuing increase in longevity and the aging of baby boomers, it is now a major public health goal to maintain optimal cognitive functioning as people age (Blazer, Yaffe, & Liverman, 2015).

Although decline in cognitive functioning is not uncommon in later life, the severity of decline is not equally distributed in the population. In particular, more educated people have better cognitive functioning in later life, experience delayed onset of cognitive pathology, and have lower risk of being diagnosed with dementia over time compared to the less educated (e.g., Clouston, Glymour, & Terrera, 2015; Langa et al., 2008; Meng & D’Arcy, 2012). However, previous research has generally investigated the association between education and cognitive functioning at the individual level and failed to consider that one’s education may have profound implications for their significant others’ cognitive functioning. Close social relationships (e.g., marital relationships) serve as a medium through which individual-level resources such as education can be translated into household-level resources that affect all family members. Recent studies reveal that spousal education is positively associated with longevity, self-rated health, and health behaviors regardless of one’s own education (Brown, Hummer, & Hayward, 2014; Monden, van Lenthe, De Graaf, & Kraaykamp, 2003; Torssander & Erikson, 2009). Yet it remains unknown whether spousal education is associated with cognitive functioning in later life.

Therefore, this study seeks to examine the link between spousal education and cognitive functioning among married older adults in the United States. Specifically, this study addresses three research questions using longitudinal couple data from the Health and Retirement Study (HRS). First, is spousal education associated with one’s level of cognitive functioning at age 65 years and the rate of change in cognitive functioning controlling for one’s own education? Second, is the association between spousal education and cognitive functioning explained by economic resources, health behaviors, spousal health status, spousal caregiving, and social activities? Third, does the association between spousal education and cognitive functioning differ for men and women?

Background

Current studies on cognitive functioning have predominantly considered education as an individual resource. This perspective suggests that education benefits cognitive functioning not only because early-life formal education exerts a direct impact on brain development but also because higher education increases access to resources that are vital to maintaining cognitive functioning throughout the life course. Such resources encompass economic resources, health-related knowledge, healthy lifestyles, and cognitively challenging occupations and leisure activities (Langa et al., 2017; Langa et al., 2008; Stern, 2012). Marriage, as one of the most significant social relationships that people enter into, facilitates an exchange of economic, psychological, and social resources between spouses. Thus, the benefits of education for cognitive functioning may extend beyond an individual and incorporate their spouse as well. From this perspective, education represents a resource that individuals may share within in the context of close social relationships such as marriage (Brown et al., 2014).

Accumulating evidence has pointed to the importance of spousal education for health regardless of one’s own education (e.g., Brown et al., 2014). Yet the study by Lee Kawachi, Berkman, and Grodstein (2003) reported that husband’s education was not significantly associated with wife’s cognitive functioning. Because this study was conducted among female nurses with at least 15 years of education, the results cannot be generalized to a broader aging population. The short follow-up period also makes it impossible to examine the long-term effect of husband’s education on women’s cognitive functioning. Thus, further investigation is needed to determine whether spousal education is associated with cognitive functioning over significant periods of later life for women, and the possible link of spousal education and men’s cognitive functioning is largely unexplored.

Conceptual Framework

More years of spousal education may benefit cognitive functioning in later adulthood by providing economic resources, promoting health behaviors, improving spousal health status, lessening the burden of caregiving for a spouse, and enhancing cognitive reserve.

Spousal education may affect cognitive functioning through its influence on economic resources at the household level. Individuals with higher levels of education are more likely to have good jobs, earn higher incomes, and less likely to experience extended periods of involuntary unemployment (Hout, 2012). Thus, individuals with a more educated spouse are likely to have higher levels of household income and less likely to experience sustained economic strain throughout the life course. By the same token, older adults whose spouse has a higher level of education are more likely to be wealthier in later life. Higher levels of household wealth can protect older adults from risk factors that compromise cognitive functioning. For instance, prior studies have shown that economic strain undermines cognitive functioning (Chiao, Botticello, & Fuh, 2014; Lynch, Kaplan, & Shema, 1997). And living in a deprived neighborhood is detrimental to cognitive functioning for older adults in spite of individual socioeconomic status (Lang et al., 2008). Wealthy older couples who can afford to live in safer and more affluent neighborhoods may avoid the detrimental effects of neighborhood deprivation on cognitive functioning.

The impact of spousal education on cognitive functioning may also work through health behaviors. Previous research shows that health behaviors, such as drinking, smoking, and body weight, significantly predict cognitive functioning (Lee et al., 2010). Individuals with higher education are more likely to adopt healthy behaviors not only because they have more health-related knowledge but also because they are better at internalizing that knowledge and adjusting their behaviors accordingly (Cutler & Lleras-Muney, 2010). When one spouse engages in healthy behaviors, he or she may attempt to monitor, regulate, and sanction the other spouse’s health behaviors to promote health, as suggested by the social control model (Umberson, 1992). Empirical evidence shows that when one spouse improves his or her health behavior the other is more likely to make change as well (Falba & Sindelar, 2008; Jackson, Steptoe, & Wardle, 2015). Consequently, spouses become increasingly similar to each other with respect to a wide range of health behaviors as shown by research on health concordance within couples (Meyler, Stimpson, & Peek, 2007). More importantly, Monden et al. (2003) show that having a more educated spouse is associated with lower risk of smoking. Thus, older adults with a more educated spouse tend to engage in more health-promoting behaviors, and numerous studies establish the link of health-promoting behaviors with reduced risk for cognitive decline (e.g., Lee et al., 2010).

Spousal health and burden of caregiving for a spouse constitute another pathway through which spousal education may affect cognitive functioning. Considerable evidence corroborates the association between early-life educational attainment and health advantages in later life (e.g., Hayward, Hummer, & Sasson, 2015). Spouses with more years of education are likely to have better health as they enter into later adulthood and are less likely to have severe limitations that require intensive caregiving in the early stages of later adulthood. Prior research shows that caregiving can be an extremely stressful experience that weakens caregivers’ physical, mental, and cognitive health (Kuzuya et al., 2011; Schulz & Sherwood, 2008; Vitaliano, Murphy, Young, Echeverria, & Borson, 2011). Thus, individuals with more educated spouses are less likely to experience the hazardous effects of spousal caregiving on cognitive functioning in later adulthood as their spouses are likely to be healthier compared to those with less educated spouses.

Another pathway through which spouse’s education may enhance cognitive functioning is cognitive reserve. Cognitive reserve refers to the ability of the brain to cope with pathology through effective use of existing cognitive processing paradigms or flexible use of alternative compensatory paradigms (Stern, 2012). Social factors that increase cognitive reserve include educational and occupational attainment as well as leisure activities in later life (Stern, 2012). People with higher education have more resources (e.g., time, money, information, and health) that allow them to engage in leisure activities (Minhat & Amin, 2012). They may also actively participate in cognitively challenging activities (e.g., reading and taking courses) as years of education have built up their abilities and confidence to deal with these challenges (Parisi et al., 2012). Cognitively challenging activities that typically involve two or more persons (e.g., discussing current issues and playing cards) may benefit not only one’s own but also one’s spouse’s cognitive functioning. Thus, individuals with more educated spouses may be exposed to more cognitively challenging activities that are conducive to cognitive functioning.

Gendered Patterns

The association between spousal education and cognitive functioning may differ for men and women. Among older cohorts, husbands’ education largely determines household income and wealth (Casper & Bianchi, 2002). Higher educational status of husbands may then contribute more, over time, to women’s cognitive functioning. In addition to the economic benefits of having a more educated spouse, women with more educated husbands may also be less likely to engage in intensive caregiving for an unhealthy spouse earlier in the life course. On the other hand, the social control model suggests that women are more likely than men to regulate and moderate their spouses’ health behaviors (Umberson, 1992). Women with more years of education are likely to adopt healthier behaviors themselves and encourage their husbands to adopt similar behaviors. Wives’ education may then be more beneficial to husbands’ cognitive functioning because of the positive effects of healthy behaviors on cognitive functioning (Lee et al., 2010). Women with higher education also have larger social networks that enable them to engage their husbands in more social activities, which ultimately benefit their husbands’ cognitive functioning (McLaughlin, Vagenas, Pachana, Begum, & Dobson, 2010; Thomas, 2011). Thus, prior studies suggest the possibility of gender differences in the association of spousal education and cognitive status, but it is unclear whether men or women would garner more cognitive benefits from higher levels of spousal education.

Taken together, the earlier discussion provides sound reasons for expecting a positive association between spousal education and cognitive functioning as well as potential gender differences in this association. Lack of measurement on cognitively simulating activities for the majority of respondents in the HRS makes it impossible to test the pathway of cognitive reserve; therefore, this study will focus only on the pathways of economic resources, health behaviors, spousal health status, spousal caregiving, and social activities. Specifically, this study tests the following hypotheses:

Hypothesis 1: Spousal education is positively associated with the level of cognitive functioning at age 65 as well as the rate of decline in cognitive functioning controlling for one’s own education.

Hypothesis 2: The association between spousal education and cognitive functioning will be significantly attenuated when economic resources, health behaviors, spousal health status, spousal caregiving, and social activities are controlled for.

Hypothesis 3: The positive association between spousal education and cognitive functioning will differ for men and women.

Method

This study uses data from the HRS, a biannual survey conducted from 1992 to the present. The initial wave was conducted among a cohort of individuals born between 1931 and 1941, and their spouses of any age. In 1998, the original HRS birth cohort was combined with several other birth cohorts to represent noninstitutionalized adults of the U.S. population older than 50 years. As cognitive functioning has been measured consistently across waves since 1996, this study used nine waves of HRS data from 1996 to 2012, provided by the RAND Center for the Study of Aging (RAND HRS Data, 2016). Spousal education was not directly measured but could be calculated for respondents whose spouse also participated in the HRS. Therefore, the analytical sample was restricted to married couples. The sample was further restricted to those having at least three records on cognitive functioning so as to estimate trajectories of cognitive functioning. Thus, each individual had his or her own baseline wave and those younger than 50 years at baseline were excluded from analysis. The final analytical sample included 5,846 individuals from 2,923 couples.

Multiple imputation was used to handle missing values for the covariates. A selection model was used to account for mortality attrition (Enders, 2010). Specifically, a discrete-time hazard model for mortality was estimated and the predicted mortality hazard was controlled in all analytical models (Xu, Thomas, & Umberson, 2016). The number of waves that respondents participated in the HRS study was also included as a control variable to account for other types of attrition (Warner & Brown, 2011).

Measures

Cognitive functioning

The cognitive measurement from the HRS includes serial subtraction of 7s, backwards counting from 20, object naming tasks, and immediate and delayed word recall of a list of 10 words. A summative score based on all these components was calculated for each individual, ranging from 0 to 35. Higher scores on this measure indicate better cognitive functioning.

Spousal education

Spousal education is measured by years of formal education completed by a respondent’s spouse, ranging from 0 to 17.

Economic resources

Economic resources are indicated by household income and household wealth. Natural logarithm transformation was applied to household income and household wealth.

Health behaviors

Alcohol consumption was coded into four dummy variables: never drank, former drinker, current light drinker (reference group), and current heavy drinker. Current light drinker was coded as 1 for individuals who consumed 7 or fewer alcoholic drinks per week (Lin, Guerrieri, & Moore, 2011). Smoking status is represented by two dummy variables: former smoker and current smoker. Individuals who never smoked served as the reference group. Body mass index (BMI) was calculated with self-reported weight and height.

Spousal health status and spousal caregiving

Spousal health status is measured by spouse’s self-rated health and functional limitations. Spouse’s self-rated health was coded as a dummy variable with 1 representing those with excellent, very good, or good health. Functional limitations were also dummy-coded. Respondents who had difficulties with any of the following activities were assigned a value of 1: bathing, eating, dressing, walking across a room, getting in or out of bed, using the telephone, taking medication, and managing money. Spousal caregiving was coded as 1 for respondents who helped their spouses with activities of daily living, instrumental activities of daily living, or money management and as 0 otherwise.

Social activities

Social activities were assessed by the total number of activities that respondents engaged in. These activities included (a) weekly or greater contact with parents, (b) weekly or greater contact with offspring, (c) weekly or greater contact with neighbors, and (d) any volunteer work for religious, educational, health-related, or other charitable organizations in the last 12 months. This measure of social activities ranged from 0 to 4.

Control variables

Respondent’s education was measured by years of schooling with values ranging from 0 to 17. Gender was coded as a dummy variable, with 1 referring to female respondents. Race was coded into three dummy variables: blacks, Hispanics, and others, with whites as the reference group. As not all couples were continuously married throughout the study period, marital transitions were accounted for by four dummy variables: divorced, widowed, spouse absent, and partnered. The number of chronic conditions that respondents had ever experienced at baseline was controlled in all models. These included high blood pressure, diabetes, cancer, lung disease, heart disease, stroke, psychiatric problems, and arthritis. To account for retest effect, the variable retest was coded as 0 for respondents’ first cognitive measurement occasion and as 1 for subsequent measurement occasions (Gross et al., 2015).

Analytic Strategy

Three-level growth curve models (GCMs) were conducted to examine the association between spousal education and cognitive functioning. Level 1 measurements of cognitive functioning are nested within Level 2 individuals who are further nested within couples at Level 3. For longitudinal couple data, the second level has only two degrees of freedom and only one random effect at the individual level can be accurately estimated (Atkins, 2005). Therefore, only the intercept at Level 2 was allowed to vary randomly and the slopes were constrained to be the same for both spouses. The equations of the GCM can be specified as:

Level 1:

$${\displaystyle \begin{array}{l}{\displaystyle Y_{tij}={\pi }_{0ij}+{\pi }_{1ij}{(\mathrm{A}\mathrm{g}\mathrm{e}-65)}_{tij}+{\pi }_{2ij}{{(\mathrm{A}\mathrm{g}\mathrm{e}-65)}_{tij}}^2+e_{tij}}\end{array}}$$

Level 2:

$${\displaystyle \begin{array}{l}{\displaystyle {\pi }_{0ij}={\mathrm{\beta }}_{00j}+{\mathrm{\beta }}_{01j}{w}_{ij}+r_{0ij}}\end{array}}$$

$$\begin{array}{l}{\pi }_{1ij}={\text{β}}_{10j}+{\text{β}}_{11j}{w}_{ij}\hfill \end{array}$$

$${\displaystyle \begin{array}{l}{\displaystyle {\pi }_{2ij}={\mathrm{\beta }}_{20j}}\end{array}}$$

Level 3:

$$\begin{array}{l}{\text{β}}_{00j}={\text{γ}}_{000}+u_{00j}\hfill \end{array}$$

$${\displaystyle \begin{array}{l}{\displaystyle {\mathrm{\beta }}_{10j}={\mathrm{\gamma }}_{100}+u_{10j}}\end{array}}$$

$${\displaystyle \begin{array}{l}{\displaystyle {\mathrm{\beta }}_{20j}={\mathrm{\gamma }}_{200}}\end{array}}$$

The first equation estimates within-individual change in cognitive functioning with the increase of age. Age was centered at 65 so that the initial status of cognitive functioning represented cognitive functioning at age 65. Y_tij represents cognitive functioning measured at age t for individual i from couple j. π_0ij is the individual intercept; π_1ij and π_2ij are the individual linear and quadratic slopes. β_00j is the average cognitive functioning at age 65 and β_10j is the average change rate in cognitive functioning at age 65 after controlling for all covariates; β_20j is the average acceleration in growth within couples. w_ij represents all the time-invariant covariates at the individual level. γ₀₀₀, γ₁₀₀, and γ₂₀₀ represent the overall average intercept, linear slope, and quadratic slope across all couples. e_tij is the within-individual error term; r_0ij is the residual for the individual intercept; u_00j and u_10j are between-couple residuals. All models were estimated using Stata 15.1.

Results

Table 1 presents the descriptive statistics for all variables included for analyses. Results from the unconditional GCM show that the average level of cognitive functioning at age 65 is 24.1880 points (p < .001). The average rate of change in cognitive functioning at age 65 is −0.1542 points (p < .001). The quadratic slope of −.0081 indicates that change in cognitive functioning accelerates at the rate of −0.0162 points as age increases by 1 year (results not shown).

Table 1.

Descriptive Statistics for Individuals, Health and Retirement Study (N = 5,846)

	Mean (SD)	N (%)
Cognitive functioning
1996	23.750 (4.615)
1998	22.707 (4.805)
2000	22.061 (4.934)
2002	22.142 (5.023)
2004	21.757 (4.957)
2006	21.734 (5.122)
2008	21.826 (5.201)
2010	21.517 (5.047)
2012	21.463 (5.110)
Spousal education	12.343 (3.198)
Respondent’s education	12.343 (3.198)
Household income (Ln)	10.609 (1.034)
Household wealth (Ln)	11.884 (2.119)
Drinking
Never drank		2,666 (45.6)
Current light drinker		1,554 (26.6)
Current heavy drinker		587 (10.0)
Former drinker		1,039 (17.8)
Smoking
Nerve smoked		2,438 (41.7)
Current smoker		689 (11.8)
Former smoker		2,719 (46.5)
BMI	27.083 (4.711)
Spousal self-rated health		4,779 (81.8)
Spousal functional limitations		617 (10.6)
Spousal caregiving		259 (4.43)
Social activities	1.841 (0.747)
Age at baseline	65.578 (8.219)
Female		2,923 (50.0)
Race
White		4,855 (83.1)
Black		509 (8.7)
Hispanic		399 (6.8)
Others		83 (1.4)
Marital duration	37.422 (13.443)
Marital transition
Divorced		32 (0.6)
Widowed		1,241 (21.2)
Spouse absent		289 (4.9)
Partnered		18 (0.3)
Chronic conditions	1.330 (1.161)
Mortality hazard	0.328 (0.275)
Number of waves	7.492 (1.774)

Note: Household income, household wealth, drinking, smoking, body mass index (BMI), spousal self-rated health, spousal functional limitations, spousal caregiving, and social activities came from the baseline wave for each individual.

Table 2 presents a series of GCMs estimating the association between spousal education and cognitive functioning. The full version of this table is available online as Supplementary Material. Model 1 shows that spousal education is positively associated with the level of cognitive functioning at age 65 and the rate of change in cognitive functioning controlling for age, race, respondent’s education, marital transitions during follow-up, and variables that account for attrition and practice effect. As spousal education increases by 1 year, the level of cognitive functioning at age 65 increases by 0.0532 points (p < .01) and cognitive decline slows down by 0.0054 points (p < .001). Figure 1 further illustrates the association of spousal education with the level of and rate of change in cognitive functioning as age increases. Respondents with high spousal education (1 SD above the mean) had better cognitive functioning at age 65 compared to those with low spousal education (1 SD below the mean). Respondents with more educated spouses experienced slower cognitive decline in later adulthood compared to those with less educated spouses.

Table 2.

Three-Level Growth Curve Models Predicting the Relationships Between Spousal Education and Cognitive Functioning, Health and Retirement Study (N = 5,846)

	Model 1	Model 2	Model 3	Model 4	Model 5
Spousal education	0.0532**	0.0189	0.0119	0.0030	0.0025
Respondent’s education	0.5559***	0.5224***	0.5106***	0.5076***	0.5018***
Household income (Ln)		0.2131***	0.1968***	0.1864***	0.1842***
Household wealth (Ln)		0.1349***	0.1246***	0.1183***	0.1138***
Never drank			−0.3886**	−0.3743**	−0.3728**
Current heavy drinker			−0.2651	−0.2779	−0.2614
Former drinker			−0.1104	−0.1065	−0.1067
Current smoker			−0.3893*	−0.3788*	−0.3444*
Former smoker			−0.0061	−0.0085	−0.0016
Body mass index			−0.0249*	−0.0242*	−0.0233*
Spousal SRH				0.3303*	0.3249*
Spousal FL				−0.7079***	−0.7002***
Spousal caregiving				0.4622	0.4761
Social activities					0.2126***
Spousal education × Age	0.0054***	0.0048**	0.0043**	0.0045**	0.0045**
Respondent’s education × Age	−0.0026	−0.0030	−0.0037*	−0.0035*	−0.0034*
Household income (Ln) × Age		0.0016	0.0010	0.0013	0.0014
Household wealth (Ln) × Age		0.0043*	0.0035	0.0037	0.0037
Never drank × Age			−0.0280**	−0.0284**	−0.0283**
Current heavy drinker × Age			0.0251	0.0256	0.0256
Former drinker × Age			0.0113	0.0113	0.0111
Current smoker × Age			−0.0225	−0.0224	−0.0230
Former smoker × Age			−0.0092	−0.0089	−0.0088
Body mass index × Age			0.0002	0.0002	0.0001
Spousal SRH × Age				−0.0032	−0.0032
Spousal FL × Age				0.0325*	0.0322*
Spousal caregiving × Age				−0.0101	−0.0104
Social activities × Age					−0.0044
Age centered at 65	−0.4200***	−0.4778***	−0.4404***	−0.4482***	−0.4486***
Age squared	−0.0094***	−0.0092***	−0.0093***	−0.0093***	−0.0093***
Constant	15.7201***	12.6631***	14.2368***	14.3068***	14.1852***
Log likelihood	−80,927.2	−80,883.6	−80,845.4	−80,833.0	−80,826.6

Note: SRH = self-rated health; FL = functional limitations. Cognitive functioning ranged from 0 to 35 points. Household income, household wealth, drinking, smoking, body mass index, spousal SRH, spousal FL, spousal caregiving, and social activities came from each individual’s baseline wave. All models adjusted for gender, race, marital duration, marital transition, chronic conditions, death hazard, number of waves, and retest effect. The full version of the table that includes all variables and the random effects is available online as supplementary material.

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

Figure 1.

Predicted trajectories of cognitive functioning by spousal education.

After adding variables on economic resources in Model 2, the association between spousal education and the level of cognitive functioning at age 65 is greatly reduced and no longer statistically significant (γ₀₀₀ = 0.0189, ns). The association between spousal education and the rate of change in cognitive functioning at age 65 is slightly reduced but remains significant (γ₁₀₀ = 0.0048, p < .01). Both household income and wealth are positively related to the level of cognitive functioning at age 65 and greater wealth is also associated with slower cognitive decline at age 65.

As shown in Model 3, introducing health behavior measures further reduces the association of spousal education with the level of cognitive functioning at age 65 (γ₀₀₀ = 0.0119, ns) and the rate of decline in cognitive functioning at age 65 (γ₁₀₀ = 0.0043, p < .01). Compared to light drinkers, those who never drank had poorer cognitive functioning and a higher rate of decline in cognitive functioning at age 65. Smoking and higher levels of BMI are associated with poorer cognitive functioning at age 65.

Model 4 introduces measures on spousal health status and spousal caregiving. The association of spousal education with the level of cognitive functioning at age 65 is further attenuated (γ₀₀₀ = 0.0030, ns); whereas the linkage between spousal education and the rate of change in cognitive functioning at age 65 increases slightly (γ₁₀₀ = 0.0045, p < .01). Self-rated health from spouse is positively associated with cognitive functioning at age 65. Spousal functional limitation is associated with worse cognitive functioning but slower cognitive decline at age 65.

Controlling for social activities in Model 5 further mitigates the association between spousal education and the level of cognitive functioning at age 65 (γ₀₀₀ = 0.0025, ns). Participation in social activities is positively associated with cognitive functioning at age 65. However, the association between spousal education and the rate of change in cognitive functioning at age 65 remains unchanged.

Further analysis of the interaction between spousal education and gender shows that the associations of spousal education with the level of cognitive functioning and the rate of change in cognitive functioning at age 65 do not vary by gender (results not shown), suggesting that higher levels of spousal education are similarly associated with cognitive functioning at later life for men and women.

Discussion

Spouses are able to pool resources to maximize health benefit; marriage also serves as a channel through which the impact of one’s education on health moves beyond each individual to include his or her spouse. This study investigates whether spousal education is associated with cognitive functioning, whether this association can be explained by the pathways of economic resources, health behaviors, spousal health status, spousal caregiving, and social activities, and whether this association differs for men and women. Results show that more years of spousal education are associated with better cognitive functioning at age 65 and this association becomes nonsignificant after controlling for economic resources. More years of spousal education are also associated with a slower rate of cognitive decline at later adulthood; this association is mitigated by economic resources and health behaviors but the association remains statistically significant. However, no gender difference in the association between spousal education and cognitive functioning has been found. These findings suggest that more years of spousal education may slow cognitive decline in later life, regardless of one’s own education and other factors known to influence cognitive decline. These findings point to the need for additional research to follow change in cognitive status over time for older couples, in relation to each spouse’s level of education.

Hypothesis 1 is supported as more years of spousal education are associated with higher level of cognitive functioning at age 65 and a slower rate of decline in cognitive functioning. This finding is consistent with previous studies on the positive association between spousal education and physical health (Brown et al., 2014; Monden et al., 2003; Torssander & Erikson, 2009). Yet this finding differs from that by Lee and colleagues (2003) suggesting that husband’s education is not associated with women’s cognitive decline. The Lee study examined only substantial cognitive decline (the worst 10% of the distribution of cognitive change) among wives within an average 2 years follow-up. A short follow-up period prevented the study from detecting the effect of husbands’ education on wives’ cognitive functioning trajectories in the long term.

Hypothesis 2 is partially supported as the present study shows that the association between spousal education and the level of cognitive functioning at age 65 can be fully explained by economic resources, suggesting that differences in cognitive functioning at age 65 by spousal education may primarily reflect disparities in economic resources. Prior research suggests that more economic resources may protect older adults from the deleterious effects of economic strain as well as neighborhood deprivation on cognitive functioning (Chiao et al., 2014; Lang et al., 2008). Higher levels of household income and wealth are also associated with more health-promoting behaviors and more leisure time to engage in socially and cognitively stimulating activities, which ultimately benefit cognitive functioning (Lee et al., 2010; Pampel, Krueger, & Denney, 2010).

In addition, the association between spousal education and the rate of change in cognitive functioning decreases slightly after controlling for economic resources and health behaviors, suggesting that economic resources and health behaviors are important pathways through which spousal education slows cognitive decline. Particularly, the positive association between household wealth and the rate of change in cognitive functioning at age 65 becomes nonsignificant after introducing health behaviors, indicating that greater wealth may be associated with slower cognitive decline because the affluent engage in healthy behaviors such as light drinking that protects them from cognitive decline (Beydoun et al., 2014).

The positive association between spousal education and the rate of cognitive decline may be further accounted for by the cognitive reserve pathway because those with more years of education are more likely to engage their spouse in cognitively stimulating activities in later life. Empirical research has found that the benefits of cognitive leisure activities on cognitive functioning are more pronounced for those with higher education perhaps because past successful experiences build competence and confidence and hence encourage individuals to participate in more challenging leisure activities (Lee & Chi, 2016). Yet those with less education may be discouraged from engaging in challenging leisure activities once they fail or deliberately avoid joining these challenging cognitive activities in the first place. Living with a more educated spouse, therefore, may increase one’s chances to participate in cognitively stimulating activities, which ultimately contributes to slower decline in cognitive functioning.

Hypothesis 3 is not supported as the positive association between spousal education and cognitive functioning is similar for husbands and wives. In the HRS cohorts, many educated women contributed less to household income and wealth than their husbands but they may have devoted more time to monitoring their husbands’ health behaviors and encouraging their husbands to participate in more social activities. Future studies should further test gender differences in the association between spousal education and cognitive functioning with younger age cohorts because the findings of this study are based on a subpopulation of older age cohorts of adults with better health.

The strengths of this study include the use of longitudinal couple data and testing of the specific pathways linking spousal education with cognitive functioning. However, several limitations should be noted. First, the relationship between spousal education and cognitive functioning may be confounded by unobserved behavioral or attitudinal traits that contribute to assortative mating. Second, the analytical sample may represent a selective group who are healthier than the general older population as only married couples with both spouses having at least three records on cognitive functioning were included for analyses. Third, spousal education is related to a very small change in the rate of cognitive decline at age 65. This small change may reflect that the measurement of cognitive functioning is subject to ceiling and floor effects and the analytical sample is a selective group with better health. Future studies should use neuropsychological measures of cognitive functioning to test whether spousal education has a stronger effect on cognitive change over time. Moreover, this study was not able to consider physical activity as a potential pathway linking spousal education to cognitive functioning because the HRS measured physical activity differently after 2002. Supplementary analysis with couples whose baseline wave started at or before 2002 suggests that vigorous exercise of 3 times a week or more over the past 12 months is associated with neither the level nor the rate of cognitive decline at age 65. Future studies should test whether finer measurement of physical activity (e.g., moderate physical activity) could account for the positive association between spousal education and the rate of change in cognitive functioning. In addition, this study did not test the pathway of cognitive reserve because the HRS did not measure cognitively stimulating activities for all respondents. Future research should aim to test whether cognitively stimulating activities could further explain the positive association between spousal education and the rate of cognitive decline in later life. Last, because this study is correlational, it is unable to parse the causal direction between spousal education and cognitive functioning.

Despite these limitations, this study advances our understanding of the association between spousal education and cognitive functioning. In particular, the significant association between spousal education and cognitive functioning net of one’s own education suggests that overlooking spousal education may underestimate health inequality associated with educational resources. Education, as a fundamental cause of health inequality, exerts not only direct influences on health but also indirect influences on health by shaping the social relationships that individuals form (Montez, Hummer, Hayward, Woo, & Rogers, 2011). The last few decades have witnessed an increase in educational assortative mating with more educated people more likely to enter into marriage and stay married (Cherlin, 2010; Schwartz & Mare, 2005). These marital trends further suggest that health disparities between the more educated and the less educated may increase in the future.

Funding

This work was supported by the grant, P2CHD042849, Population Research Center and the grant, T32HD007081, Training Program in Population Studies, awarded to the Population Research Center at University of Texas at Austin by the Eunice Kennedy Shriver National Institute of Child Health and Human Development.

Conflict of Interest

The authors declared no conflict of interest.