Lifespan Socioeconomic Context: Associations With Cognitive Functioning in Later Life

Lifespan Socioeconomic Context: Associations With Cognitive Functioning in Later Life Abstract Objectives Early socioeconomic status (SES) correlates with later-life cognition. However, the effect of socioeconomic context (SEC), which reflects influences from broader ecological contexts, has not been examined. The present study developed a measure of SEC using lifetime residential addresses and examined SEC and residential mobility effects on later-life cognition. Method Older adults (N = 117, Mage = 75) reported addresses since birth. Latent SEC was constructed from census income, employment, and education (1920–2010) for each county and census year, extrapolated between census years. Controlling for current SES, SEC in childhood (ages 0–18) and adulthood (ages 19–60), with finer granulations in young adulthood (ages 19–39) and midlife (ages 40–60), predicted later-life cognition. Effects of residential mobility on later-life cognition were also examined. Results Higher childhood and adulthood SEC were associated with better Auditory Verbal Learning Test recognition (β = .24, p = .012) and immediate recall (β = .26, p = .008). Higher midlife SEC was associated with faster task switching (β = .26, p = .025) and better task switching efficiency (β = .27, p = .022). Higher residential mobility in childhood was associated with higher crystallized intelligence (β = .194, p = .040). Discussion Independent of current SES, childhood and adulthood SEC predicted later-life cognition, which may be sensitive to effects of social institutions and environmental health. SEC assessed across the lifespan, and related residential mobility information may be important complements to SES in predicting later-life cognitive health. Development, Early life, Executive functions, IQ, Memory, Socioeconomic One of the best-studied individual influences on the degree and rate of cognitive development is socioeconomic status (SES), defined as an individual’s social position in relation to others and often operationalized as some combination of income, education, and occupation (Krieger, 2001; Krieger, Williams, & Moss, 1997). Development, however, occurs in context. Ecological systems theory (Bronfenbrenner, 1977) embeds the individual in a close social context (microsystem), a broader community context (exosystem), and an even broader cultural context (macrosystem), all of which interact with each other and all of which can influence development and health, including cognitive health. Socioeconomic context (SEC) is an exosystem-level variable that reflects the relative wealth and quality of an individual’s community, which in turn correlates with factors such as environmental exposures, educational quality, and access to health care. SEC and SES can be operationalized using similar variables at different levels of analysis (e.g., county- vs individual-level income, education, and unemployment rate or occupation information). However, SEC is not simply an expanded measure of SES but captures emergent properties at broader contextual levels (not always reflected at the individual level) that arise from interactions among historical, environmental, and community circumstances (G. W. Evans & Kantrowitz, 2002; Krieger et al., 2008; McLeod & Kessler, 1990; Yang, Gerken, Schorpp, Boen, & Harris, 2017). For example, in rural communities, local culture, economy, and geographic location determine health disparities via local tax base size, poverty rates, and access to education and health care (Thomas, DiClemente, & Snell, 2014). The importance of SEC is emphasized by Bronfenbrenner’s (1977) proposition that “research on the ecology of human development requires investigations that go beyond the immediate setting containing the person to examine the larger contexts, both formal and informal, that affect events within the immediate setting” (p. 527). The overlap between SES and SEC may explain the relationship between family income and environmental exposures including air pollutants, water quality, and ambient noise, as well as the quality of residences, educational facilities, and neighborhoods (G. W. Evans & Kantrowitz, 2002). However, SEC effects may trump SES effects; in the United States, the increased risk for premature mortality associated with low SES is ameliorated for people who lived in high SEC environments (Krieger et al., 2008). Thus, SEC captures resources and exposures that are not necessarily reflected at the individual level. SEC may affect cognitive development and result in differences in neuropsychological performance in older age. Cognitive and brain development may be negatively affected by some environmental exposures associated with SEC such as environmental toxins and pollutants (Lanphear, 2015). In contrast, exposure to green space (e.g., public parks) during childhood and adulthood may support healthy cognitive aging in later life (Cherrie et al., 2018). Other exposures may affect the brain because they are stressful (e.g., noise, crowding). Stress is associated with higher systemic inflammation, which, in turn, may compromise cognition and brain health, particularly in early life (Lupien, McEwen, Gunnar, & Heim, 2009; Miller, Chen, & Parker, 2011; Wright et al., 2006). At the same time, residing in potentially more stimulating urban (vs rural) environments, particularly during childhood, is associated with higher global and executive cognitive function (Cassarino, O’Sullivan, Kenny, & Setti, 2016). Childhood may be a critical period for SEC exposure. Less socioeconomic hardship in this developmental stage may increase the probability of successful aging, defined as being relatively free from health and psychological problems in older age (Berkman et al., 1993). Better home and neighborhood environmental conditions (e.g., sanitary infrastructure) and greater concentrations of affluent neighbors are associated with better cognition in children (Duncan, Brooks-Gunn, & Klebanov, 1994; Santos et al., 2008). These early life cognitive differences may set people on certain trajectories that carry into older age. Childhood SEC may also provide certain protective resources, over and above SES, on later-life cognition. For example, among older adults who report lower childhood SES, those that live in relatively affluent urban neighborhoods fare better cognitively than their counterparts in more disadvantaged urban neighborhoods (Wight et al., 2006). Although the current study focuses on SEC, we recognize that individual-level SES is an important determinant of adult cognition. Cross-sectionally and longitudinally, lower SES is associated with worse cognition and higher risk of cognitive aging disorders (Barnes, Mendes de Leon, Wilson, Bienias, & Evans, 2004; Fotenos, Mintun, Snyder, Morris, & Buckner, 2008), whereas higher SES is protective and associated with lower risk of cognitive decline, including Alzheimer’s disease (D. A. Evans et al., 1997; Hackman, Farah, & Meaney, 2010; Jiang, Yu, Tian, & Tan, 2013; Sattler, Toro, Schönknecht, & Schröder, 2012). Additionally, higher SES in childhood is associated with better cognition function in older age (Kaplan et al., 2001; Luo & White, 2005). This investigation is the first to empirically test whether childhood and adulthood SEC contribute to later-life cognition above and beyond current individual-level SES and, in additional post hoc sensitivity analyses, above and beyond broad proxies of childhood SES. Assessment of Lifetime SEC A challenge in assessing lifetime socioeconomic conditions among older adults, whether individual status or community context, is reliance on retrospective self-reports. First, older adults must remember increasingly temporally distant information to provide a picture of earlier SES or SEC, potentially resulting in less accurate reports. Second, older adults typically have positive cognitive biases that may affect the accuracy of their memories of people and situations (Charles, Mather, & Carstensen, 2003). Specifically, in the positivity effect, focus shifts from negative information in youth to the reciprocal with age such that older adults’ memories are more likely biased toward positive information (Carstensen, 2006). For example, compared with objective data, older adults self-reported less childhood adversity (Maughan & Rutter, 1997), suggesting that they may be more forgetful of negative autobiographical events or have a positively biased memory for childhood experiences or adverse conditions. Inaccuracies in older adults’ recollections of subjective socioeconomic proxies can compromise attempts to link environments in childhood to health in older age. As an illustration, older men’s self-reported recall of childhood SES was not associated with premature mortality or coronary risk, but social disadvantage assessed from objective data including school health records was robustly associated with higher all-cause mortality risk (odds ratio [OR] = 1.41) and acute coronary event risk (OR = 1.50; Kauhanen et al., 2006). In the present study, older adults reported their lifetime residential addresses in an attempt to generate a less subjective and potentially more accurate measure of lifetime SEC. Objective information such as addresses are less subject to positive memory biases and may be recalled with fewer errors than subjective measures such as individual perceptions of socioeconomic proxies (Straughen, Caldwell, Osypuk, Helmkamp, & Misra, 2013; Ward, 2011). By linking these addresses to county-level data from the U.S. Census, participants’ residences were linked to the broad SEC of each county at particular points in time. Reports of lifetime residential addresses also permitted examination of residential mobility, the frequency with which an individual changes residential location in a specified time period. The frequency of moves may be especially helpful in investigating a dose–response effect of stressors associated with residential change. Higher residential mobility in childhood and adulthood is associated with poverty, unemployment, and family disruption (Jelleyman & Spencer, 2007). The Present Study The present investigation represents the first use of lifetime residential addresses to investigate effects of lifetime SEC and residential mobility on later-life cognition. We hypothesized that higher SEC, particularly in early life, would be associated with better later-life cognition, above and beyond current individual-level SES. We also hypothesized that less residential mobility would be associated with the better cognitive function. To provide a broad assessment of cognition in domains that may be vulnerable to socioeconomic influences, several neuropsychological functions were assessed, including estimated crystallized intelligence, visuomotor speed and task switching speed and efficiency, and verbal memory (verbal learning, immediate recall, delayed recall, and recognition). Most of the cohort (97%) lived in the same county at the time of neuropsychological testing, with 3% living in adjacent communities, providing equifinality with regard to SEC. Therefore, the present study tested the effects of lifetime county-level SEC on later-life cognition where current county-level SEC was constant across the cohort. Finally, to explore potential SEC effects specific to developmental periods, we tested associations between childhood SEC, young adulthood SEC, and midlife SEC on later-life cognition. Method Participants Participants were 117 community-dwelling older adults over the age of 60 (Mage= 75 years, range = 60–92) participating in an ongoing longitudinal study of psychological, cognitive, and immunological health. The sample was 60% female, consistent with the general older adult population according to the 2010 U.S. Census (57% female aged 65 years and older; American Fact Finder, 2017). A majority (93%) of the sample was Caucasian, with the remainder African American. Median education was 16 years (range = 9–22 years), and median household income was $78,500 (range = $9,000–$400,000). Because the study involved measurement of immune parameters, exclusion criteria were: diseases or disorders affecting the immune system; radiation therapy or chemotherapy within 5 years; unwillingness to undergo venipuncture; or more than two medications in the classes of psychotropics, antihypertensives, hormone replacement, or thyroid supplements. Additionally, use of any of the following medications was exclusionary: medications for cognitive impairment (e.g., donepezil, memantine, rivastigmine), opiates, systemic steroids, cytotoxic drugs, and tumor necrosis factor blockers. Thus, participants represent a generally healthy community sample of older adults. The majority of the sample scored within normal limits on the neuropsychological measures. Eight individuals scored in the clinical range (Z < −2.0) on 1–3 parameters of the Auditory Verbal Learning Test (AVLT), but their scores on the other AVLT parameters and the Trail Making Test were in the normal range. Design and Procedures Participants were recruited from a prospective research pool maintained by the Sanders-Brown Center on Aging at the University of Kentucky. In the longitudinal study, participants are interviewed, undergo neuropsychological testing, and have blood drawn every 6 months. The present study utilized only the neuropsychological assessments performed at the initial assessment, when participants provided lifetime residential addresses, to reduce the influence of practice effects. This study was conducted with the approval of the University of Kentucky Institutional Review Board, and all participants provided informed consent for study procedures. Measures Socioeconomic context SEC was measured using self-reported addresses from birth until age 60. Addresses were coded for county-level information using the publicly accessible database Social Explorer (https://www.socialexplorer.com), which provides access to U.S. Census Bureau data from 1790 to 2010. In the event a participant could not remember an address, they were asked if they could provide a specific location or landmark near that residence (e.g., township name, cross streets, nearby hospital or school) to enable researchers to identify the county of residence. Data were missing (details below) if participants could not recall either an address or usable landmark for a particular residence. For each county of residence reported in the sample, the percent unemployed, percent with at least a high school diploma, and/or median household income (in 2010 dollars) were obtained for each census year from 1920 to 2010. The percent unemployed was available for decades 1930–1950 and 1970–2010; percent with at least a high school diploma, 1920–1950 and 1970–2010; and median household income, 1970–2010. Data for the 1960 Census were not yet analyzed by the U.S. Census Bureau and therefore not available. Values for percent unemployed and percent with at least a high school diploma were squared to normalize their distributions. Variables were standardized (M = 0, SD = 1) within census years to remove between-census variation (i.e., progressive increases in education levels over the century) while maintaining between-county relationships. Education, income, and employment values from all counties and census years were used to construct an SEC latent variable in Mplus; scores were output for each county and census year. Noncensus year scores were generated by extrapolating between census years for each county. The chi-square test was statistically significant, as is often true with large sample sizes [χ2(3) = 985.44, p < .001], but fit indices indicated that the model fits well (comparative fit index = 1.00; Tucker–Lewis index = 1.00; AIC = 14771.3; BIC = 14822.6). All three indicators had statistically significant loadings on the SEC latent variable (education, β = .86, p < .001; income, β = .78, p < .001; employment, β = .30, p < .001). To account for missing SEC data (5.4% person-year observations) due to the overseas residence and active military duty (1.9%) and failure to recall a residence (3.5%), we implemented multiple imputation using the Amelia package (version 1.7.3) in R (version 3.0.3). This method implements an expectation–maximization with bootstrapping algorithm to impute missing values. The variables used to generate imputed scores were person, year, SEC, and county. A model including a linear effect of time was used to generate imputed data sets. SEC scores from five imputed data sets were averaged together to obtain the final SEC scores. Participants were assigned the relevant SEC value for each year of their life from birth until age 60 based on the county in which they lived at each age. Figure 1 illustrates SEC data through age 60 for five selected participants. Childhood SEC was calculated by averaging the SEC values for each participant from birth through age 18. Adulthood SEC was calculated by averaging the SEC values for ages 19 through 60. In addition, for exploratory analyses, adulthood SEC was subdivided into young adulthood SEC (ages 19–39) and midlife SEC (ages 40–60), which were calculated in the same manner. Figure 1. View largeDownload slide Example socioeconomic context (SEC) trajectories. This figure illustrates SEC scores for five representative individuals through age 60. Higher SEC values correspond to higher wealth and quality of an individual’s community. Figure 1. View largeDownload slide Example socioeconomic context (SEC) trajectories. This figure illustrates SEC scores for five representative individuals through age 60. Higher SEC values correspond to higher wealth and quality of an individual’s community. Residential mobility Using the lifetime address reports, residential mobility was calculated as the number of moves a participant made, including moves within the same county and to a different county. Residential mobility was calculated for childhood (ages 0–18) and adulthood (ages 19–60). Socioeconomic status To control for the effect of current individual SES in SEC analyses, the Hollingshead Index was used as a multidimensional composite encompassing educational attainment, occupational prestige, sex, and marital status (Hollingshead, 1975). This method assumes that the education and occupation of both spouses play an equal part in the SES of the family. Participants reported their age, sex, current household income, their own and their spouses’ (if married) education and occupations, or previous occupations if retired. A. B. Scott coded educational attainment and occupational prestige. Educational attainment was coded on a 7-point scale (i.e., 7 = graduate/professional training; 1 = less than 7th grade). Occupational prestige, defined as the level of assumed skill and power individuals possess in the maintenance functions of their position, was coded on a 9-point scale (e.g., 9 = higher executive, proprietor of large business, major professional; 1 = farm laborers, menial service workers). Following methods outlined by Hollingshead (1975), education and occupation scores were multiplied by factor weights (3 and 5, respectively). The SES score for a married participant was then calculated by summing the education and occupation scores for the participant and their spouse and dividing by two. The SES score for an unmarried participant was calculated by summing the individual’s education and occupation scores. Crystallized intelligence The North American Adult Reading Test (NAART; Blair & Spreen, 1989) was used to estimate crystallized intelligence. The NAART is a reliable and valid estimate of premorbid intelligence, comparable to the Wechsler Adult Intelligence Scale-Revised. The NAART is thought to be relatively resistant to neurological damage, though the change may occur under certain contexts. Decline may occur for those with mild to severe dementia, particularly when language is comprised. Factors such as cognitive reserve, social class, and education may also ameliorate decline (Strauss, Sherman, & Spreen, 2006). The NAART requires participants to read aloud a list of 61 irregularly spelled words; participants are scored for correct pronunciation. The NAART provides regression equations that estimate full-scale IQ (FSIQ), verbal IQ, and performance IQ based on the number of pronunciation errors (Blair & Spreen, 1989). The FSIQ estimate was utilized in the present study. Visuomotor speed and task switching The Trail Making Test (TMT) parts A and B were used to assess visuomotor speed and task switching speed, respectively (Reitan, 1955). Additionally, task switching efficiency was assessed using the derived TMT Part B minus A index (Drane, Yuspeh, Huthwaite, & Klingler, 2002). In Part A, participants are timed while drawing lines connecting numbers in ascending order as quickly as possible (i.e., 1-2-3, etc.). Similarly, in Part B, participants draw lines switching between numbers and letters in ascending and alphabetical order (i.e., 1-A-2-B-3-C, etc.). The TMT taps visuomotor speed in Part A and adds an executive component, task switching, in Part B, as flexibly switching between numbers and letters is required. However, as both parts A and B rely on visuomotor speed, the difference in completion time between the two parts (B minus A) may be used as a derived index to isolate the executive component of Part B, termed task switching efficiency. Normed scores based on age were used (Drane et al., 2002). Verbal memory The AVLT (Strauss et al., 2006) assessed several components of verbal memory including verbal learning, immediate recall, delayed recall, and recognition memory. The AVLT includes a 15-word list (List A), which was read to participants five times; participants were asked to recall as many words as possible after each trial (verbal learning; sum of A1–A5). Next, an interference list of 15 different words (List B) was read, and participants were asked to recall as many words as possible. After the recall of List B, participants were asked to recall words from List A (immediate recall; A6). Delayed recall of List A was assessed 20 min after immediate recall (delayed recall; A7), followed by a recognition trial including both target (recognition memory; List A) and distractor words (Strauss et al., 2006). Normed scores based on age were used (Ivnik et al., 1990). Data Analysis To describe SEC trajectories, multilevel models were used with years at Level 1 and people at Level 2 in SAS (9.4) PROC MIXED. Null models with restricted maximum likelihood estimation provided estimates of variance components and the intraclass correlation (ICC). Higher ICC values indicate more of the variance is due to stable between-person individual differences and less is due to within-person change over time. Growth models with maximum likelihood estimation and between-within degrees of freedom included the linear and quadratic effects of age as fixed effects. Fixed effects of age are reported as gamma weights, which are analogous to unstandardized beta weights in regression. The Level 1 equation is presented below: SECij=β0j+β1j(Agei) +β2j(Age2i) +εij SEC trajectories for person j at year i are a function of an intercept (SEC at the first time point), a linear and quadratic age slope, and a within-person residual. Random effects were included when indicated by the likelihood ratio test with mixture degrees of freedom. Specifically, random effects of intercept and linear age were included, but the quadratic effect of age was uniform across the sample and therefore not included. The Level 2 equation is presented below: β0j=γ00+ U0j β1j=γ10+ U1j β2j=γ20 Thus, with substitution, this model can be written as: SECij=γ00+ U0j+ Age (γ10+ U1j) + Age2(γ20) +εij To test our hypotheses, we employed two sets of hierarchical regressions in SPSS (24.0). Specifically, we regressed the neurocognitive variables on (1) SEC in childhood and adulthood, controlling for current SES, and (2) residential mobility in childhood and adulthood. A third set of exploratory hierarchical regressions evaluated associations between SEC in childhood and in developmental periods within adulthood, namely young adulthood (ages 19–39) and midlife (ages 40–60), on later-life cognition. Last, we conducted post hoc sensitivity analyses that further controlled for childhood SES in the SEC analyses. Two items from the Childhood Trauma Questionnaire (CTQ; Bernstein et al., 2003) were included to broadly account for childhood material wealth. The two items were “I didn’t have enough to eat” and “I had to wear dirty clothes”; participants responded on a 5-point scale from “never true” to “often true”. In general, this sample did not have severe deprivation in childhood; the majority of participants endorsed “never true” for both items (not enough to eat: 80%; wore dirty clothes: 82%), and the remaining endorsed “rarely true” (not enough to eat: 6%; wore dirty clothes: 5%), “sometimes true” (not enough to eat: 4%; wore dirty clothes: 4%); and “often true” (not enough to eat: 1%; wore dirty clothes: 0%). CTQ data for 11 participants (9%) were missing. Because some participants (N = 56) were administered the TMT and AVLT in an earlier study phase, a dummy variable was included to account for practice effects (1 = previously tested; 0 = not previously tested). Last, we did not include age in the models because crystallized intelligence remains largely intact into old age (e.g., Christensen, 2001) and because the neurocognitive scores for visuomotor speed and task switching (TMT) and verbal memory (AVLT) were already normed based on age. The Benjamini and Hochberg (1995) procedure was applied to the results to correct for multiple comparisons. Significance levels correcting for the false discovery rate (FDR) of 0.10 were calculated (a) for each hierarchical model/step and (b) for each predictor across hierarchical models separated into three families of outcomes, representing crystallized intelligence, visuomotor processing (three variables: visuomotor speed, task switching, and task switching efficiency), and verbal memory (four variables: total learning, immediate recall, delayed recall, and recognition). Results Lifetime Residential Addresses: Descriptive Statistics The majority of lifetime residential addresses (63% person-year observations) were in the Commonwealth of Kentucky. Within Kentucky, there were 46 different counties represented, with variability in the SEC values across these counties (M = 0.25, SD = 0.58, range = −2.81 to 0.99). The remaining addresses (37% person-year observations) were from 198 different counties outside of Kentucky, which had slightly more SEC variability (M = 0.32, SD = 0.67, range = −1.83 to 3.10). Table 1 depicts descriptive statistics and correlations among demographic data, SEC and SES parameters, and residential mobility. SEC was significantly lower [t(113)= −7.38, p < .001] in childhood (M = −0.048, SD = 0.66) than adulthood (M = 0.42, SD = 0.29). In exploratory models, we also examined SEC in young adulthood (19–39 years) and midlife (40–60 years). SEC was significantly lower [t(113)= −2.56, p = 0.012) in young adulthood (M = 0.38, SD = 0.37) than in midlife (M = 0.46, SD = 0.30). Most variation in lifetime (birth to age 60) SEC was due to within-person changes (ICC = .25). Adulthood SEC (ICC = .35) had more within-person variability than childhood SEC (ICC = .76), which was more stable. Additionally, young adulthood SEC had more within-person variability (ICC = .40) than midlife SEC (ICC = .60). Growth models indicated that SEC increased slightly with age (linear γ = .01, SE = .002, p< .001), with slower increases in older age (quadratic γ= −.0002, SE = .00002, p < .001). A statistically significant (p < .001) random effect of age indicated that there were individual differences in SEC trajectories. However, examination of Figure 1 suggests that most of the variation in SEC was due to nonlinear changes in the level of exposure over time rather than linear trends across the lifespan. On average, participants moved twice (SD = 1.97, range = 0–9 moves) during childhood (ages 0–18) and seven times (SD = 3.02, range = 0–16) during adulthood (ages 19–60 years). Table 1. Means of and Correlations Between Age, Gender, SEC, SES, and Residential Mobility Variable Mean (SD) Correlations 1 2 3 4 5 6 7 8 9 10 11 1. Age 75 (5.9) 2. Gender 60% female −.06 3. Childhood SEC −0.06 (0.66) −.01 .17 4. Adulthood SEC 0.42 (0.29) .15 −.06 .19* 5. Young adulthood SEC 0.38 (0.37) .14 .01 .35** .90** 6. Middle adulthood SEC 0.46 (0.30) .13 −.12 −.07 .84** .51** 7. Childhood SES (enough to eat) 0.19 (0.55) .08 −.07 −.11 .07 .04 .09 8. Childhood SES (wore dirty clothes) 0.13 (0.43) .17 −.15 −.02 −.11 −.13 −.06 .33** 9. Current SES 55.2 (7.9) .10 .01 −.04 −.13 −.11 −.11 −.08 .04 10. Moves in childhood 2 (2.0) −.05 .05 .04 .13 .16 .05 .13 .04 .07 11. Moves in adulthood 7 (3.0) .24** .03 −.02 −.10 −.13 −.03 .10 .22* .19* −.05 12. Total moves 9 (3.6) .17 .02 .02 −.02 −.03 .01 .15 .20* .18* .52** .83** Variable Mean (SD) Correlations 1 2 3 4 5 6 7 8 9 10 11 1. Age 75 (5.9) 2. Gender 60% female −.06 3. Childhood SEC −0.06 (0.66) −.01 .17 4. Adulthood SEC 0.42 (0.29) .15 −.06 .19* 5. Young adulthood SEC 0.38 (0.37) .14 .01 .35** .90** 6. Middle adulthood SEC 0.46 (0.30) .13 −.12 −.07 .84** .51** 7. Childhood SES (enough to eat) 0.19 (0.55) .08 −.07 −.11 .07 .04 .09 8. Childhood SES (wore dirty clothes) 0.13 (0.43) .17 −.15 −.02 −.11 −.13 −.06 .33** 9. Current SES 55.2 (7.9) .10 .01 −.04 −.13 −.11 −.11 −.08 .04 10. Moves in childhood 2 (2.0) −.05 .05 .04 .13 .16 .05 .13 .04 .07 11. Moves in adulthood 7 (3.0) .24** .03 −.02 −.10 −.13 −.03 .10 .22* .19* −.05 12. Total moves 9 (3.6) .17 .02 .02 −.02 −.03 .01 .15 .20* .18* .52** .83** Notes: Gender is coded 1 = male, 2 = female. Childhood SEC (ages 0–18); adulthood SEC (ages 19–60); young adulthood SEC (ages 19–39); middle adulthood SEC (ages 40–60). SEC = socioeconomic context; SES = socioeconomic status. *p < .05. **p < .01. View Large Table 1. Means of and Correlations Between Age, Gender, SEC, SES, and Residential Mobility Variable Mean (SD) Correlations 1 2 3 4 5 6 7 8 9 10 11 1. Age 75 (5.9) 2. Gender 60% female −.06 3. Childhood SEC −0.06 (0.66) −.01 .17 4. Adulthood SEC 0.42 (0.29) .15 −.06 .19* 5. Young adulthood SEC 0.38 (0.37) .14 .01 .35** .90** 6. Middle adulthood SEC 0.46 (0.30) .13 −.12 −.07 .84** .51** 7. Childhood SES (enough to eat) 0.19 (0.55) .08 −.07 −.11 .07 .04 .09 8. Childhood SES (wore dirty clothes) 0.13 (0.43) .17 −.15 −.02 −.11 −.13 −.06 .33** 9. Current SES 55.2 (7.9) .10 .01 −.04 −.13 −.11 −.11 −.08 .04 10. Moves in childhood 2 (2.0) −.05 .05 .04 .13 .16 .05 .13 .04 .07 11. Moves in adulthood 7 (3.0) .24** .03 −.02 −.10 −.13 −.03 .10 .22* .19* −.05 12. Total moves 9 (3.6) .17 .02 .02 −.02 −.03 .01 .15 .20* .18* .52** .83** Variable Mean (SD) Correlations 1 2 3 4 5 6 7 8 9 10 11 1. Age 75 (5.9) 2. Gender 60% female −.06 3. Childhood SEC −0.06 (0.66) −.01 .17 4. Adulthood SEC 0.42 (0.29) .15 −.06 .19* 5. Young adulthood SEC 0.38 (0.37) .14 .01 .35** .90** 6. Middle adulthood SEC 0.46 (0.30) .13 −.12 −.07 .84** .51** 7. Childhood SES (enough to eat) 0.19 (0.55) .08 −.07 −.11 .07 .04 .09 8. Childhood SES (wore dirty clothes) 0.13 (0.43) .17 −.15 −.02 −.11 −.13 −.06 .33** 9. Current SES 55.2 (7.9) .10 .01 −.04 −.13 −.11 −.11 −.08 .04 10. Moves in childhood 2 (2.0) −.05 .05 .04 .13 .16 .05 .13 .04 .07 11. Moves in adulthood 7 (3.0) .24** .03 −.02 −.10 −.13 −.03 .10 .22* .19* −.05 12. Total moves 9 (3.6) .17 .02 .02 −.02 −.03 .01 .15 .20* .18* .52** .83** Notes: Gender is coded 1 = male, 2 = female. Childhood SEC (ages 0–18); adulthood SEC (ages 19–60); young adulthood SEC (ages 19–39); middle adulthood SEC (ages 40–60). SEC = socioeconomic context; SES = socioeconomic status. *p < .05. **p < .01. View Large Childhood and adulthood SEC were modestly correlated (r = .19, p = .042); however, childhood SEC was related to young adulthood SEC (r = .35, p < .001) but not midlife SEC (r = −.07, p = .47). Also note that SES and SEC, whose distinct effects we are examining, were uncorrelated (current SES and adulthood SEC: r = −.13, p = .17; current SES and childhood SEC: r = −.04, p = .72; childhood SES and childhood SEC: rs = −.11 to −.02, ps = .25−.82). Childhood and adulthood residential mobility were uncorrelated (r = −.05, p = .61). Residential mobility in adulthood was positively correlated with age (r = .24, p = .009). SEC and residential mobility were not strongly related, but higher current SES was modestly correlated with more residential mobility in adulthood (r = .18, p = .049). Socioeconomic Context and Cognition Regarding the influences of broad socioeconomic environmental conditions, we hypothesized that higher SEC, particularly in the early developmental years, would be associated with better later-life cognition. Analyses included current SES in the final step of hierarchical regression models to examine whether SEC effects remained above and beyond current SES. Table 2 presents hierarchical regression results. In line with our hypothesis, higher childhood SEC was associated with better recognition memory (β = .240, p = .012). This effect remained when childhood SEC was the only main effect (Step 1), when adulthood SEC was added (Step 2), and when current SES was also added (Step 3). Higher adulthood SEC was significantly associated with better immediate recall (β = .259, p = .008) above and beyond current SES. Table 2. Hierarchical Regression Models of Child and Adult SEC on Late-Life Cognition Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 Β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027† .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a,b Step 2 .001 .029† .032† .020 .006 .065* .023 .005  Practice .164 .121 .087 .108 .052 .175 −.037  Child SEC .159 −.061 −.127 −.129 .079 .058 .000 .230*a,b  Adult SEC .030 .175† .182† .143 .080 .260*a,b .156 .073 Step 3 .048* .001 .000 .000 .001 .000 .001 .035*  Practice .170 .120 .084 .114 .051 .181 .003  Child SEC .161† −.060 −.128 −.130 .078 .058 −.001 .230*b  Adult SEC .058 .180† .181† .141 .085 .259*a,b .160 .102  Current SES .220*a,b .037 −.010 −.018 .031 −.005 .028 .192* Total R2 .076* .055 .056 .039 .024 .076 .051 .100* n 112 112 110 110 110 110 110 110 Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 Β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027† .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a,b Step 2 .001 .029† .032† .020 .006 .065* .023 .005  Practice .164 .121 .087 .108 .052 .175 −.037  Child SEC .159 −.061 −.127 −.129 .079 .058 .000 .230*a,b  Adult SEC .030 .175† .182† .143 .080 .260*a,b .156 .073 Step 3 .048* .001 .000 .000 .001 .000 .001 .035*  Practice .170 .120 .084 .114 .051 .181 .003  Child SEC .161† −.060 −.128 −.130 .078 .058 −.001 .230*b  Adult SEC .058 .180† .181† .141 .085 .259*a,b .160 .102  Current SES .220*a,b .037 −.010 −.018 .031 −.005 .028 .192* Total R2 .076* .055 .056 .039 .024 .076 .051 .100* n 112 112 110 110 110 110 110 110 Note: FDR = false discovery rate; SEC = socioeconomic context; SES = socioeconomic status. aStatistically significant after FDR correction applied within each hierarchical step. bStatistically significant after FDR correction applied within each predictor. †p < .10. *p < .05. View Large Table 2. Hierarchical Regression Models of Child and Adult SEC on Late-Life Cognition Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 Β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027† .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a,b Step 2 .001 .029† .032† .020 .006 .065* .023 .005  Practice .164 .121 .087 .108 .052 .175 −.037  Child SEC .159 −.061 −.127 −.129 .079 .058 .000 .230*a,b  Adult SEC .030 .175† .182† .143 .080 .260*a,b .156 .073 Step 3 .048* .001 .000 .000 .001 .000 .001 .035*  Practice .170 .120 .084 .114 .051 .181 .003  Child SEC .161† −.060 −.128 −.130 .078 .058 −.001 .230*b  Adult SEC .058 .180† .181† .141 .085 .259*a,b .160 .102  Current SES .220*a,b .037 −.010 −.018 .031 −.005 .028 .192* Total R2 .076* .055 .056 .039 .024 .076 .051 .100* n 112 112 110 110 110 110 110 110 Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 Β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027† .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a,b Step 2 .001 .029† .032† .020 .006 .065* .023 .005  Practice .164 .121 .087 .108 .052 .175 −.037  Child SEC .159 −.061 −.127 −.129 .079 .058 .000 .230*a,b  Adult SEC .030 .175† .182† .143 .080 .260*a,b .156 .073 Step 3 .048* .001 .000 .000 .001 .000 .001 .035*  Practice .170 .120 .084 .114 .051 .181 .003  Child SEC .161† −.060 −.128 −.130 .078 .058 −.001 .230*b  Adult SEC .058 .180† .181† .141 .085 .259*a,b .160 .102  Current SES .220*a,b .037 −.010 −.018 .031 −.005 .028 .192* Total R2 .076* .055 .056 .039 .024 .076 .051 .100* n 112 112 110 110 110 110 110 110 Note: FDR = false discovery rate; SEC = socioeconomic context; SES = socioeconomic status. aStatistically significant after FDR correction applied within each hierarchical step. bStatistically significant after FDR correction applied within each predictor. †p < .10. *p < .05. View Large Turning to the influences of current individual socioeconomic position, higher current SES was significantly associated with higher estimated crystallized intelligence (β = .220, p = .019) and tended to be associated with better recognition memory (β = .192, p = .045), however this effect did not pass FDR corrections. These SES effects were above and beyond childhood and adulthood SEC effects. There were no significant associations between childhood SEC, adulthood SEC, or current SES with visuomotor speed, task switching speed or efficiency, verbal learning, or delayed recall. Higher childhood SEC tended to be associated with higher estimated crystallized intelligence (β = .161, p = .088). Additionally, higher adulthood SEC tended to be associated with better visuomotor speed (β = .180, p = .065) and faster task switching (β = .181, p = .066), however, these associations were not statistically significant. Exploratory analyses (Table 3) evaluated associations between SEC in childhood and in developmental periods within adulthood, namely young adulthood (ages 19–39) and midlife (ages 40–60), on later-life cognition. The aforementioned significant effects (Table 2) remained with the exception that associations between childhood SEC and recognition memory did not pass FDR corrections in Steps 2–4. Additional effects also emerged: when including childhood SEC and young adulthood SEC in the model, higher midlife SEC was significantly associated with faster task switching (β = .262, p = .025) and better task switching efficiency (β = .268, p = .022). These results remained significant when current SES was included in the model (Step 4) and after FDR correction applied within each predictor. Additionally, higher young adulthood SEC was associated with better immediate recall (β = .265, p = .009). This effect decreased and became nonsignificant in Steps 3 (including midlife SEC, β = .214, p = .082) and 4 (including current SES, β = .213, p = .085). Young adulthood SEC was not significantly associated with any other later-life cognition variables. Table 3. Hierarchical Regression Models of Child, Young Adult (YA), and Middle Adult (MA) SEC on Later-Life Cognition Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027 .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a Step 2 .005 .022 .010 .003 .009 .062* .024 .003  Practice .157 .111 .079 .106 .043 .170 −.040  Child SEC .139 −.085 −.132 −.124 .059 .013 −.029 .221*  YA SEC .074 .160 .106 .058 .099 .265*a .164 .056 Step 3 .006 .008 .045* .047* .000 .005 .001 .003  Practice .165 .133 .101 .105 .049 .173 −.035  Child SEC .111 −.054 −.054 −.044 .054 .038 −.018 .242*  YA SEC .133 .095 −.054 −.106 .109 .214† .141 .012  MA SEC −.096 .107 .262*b .268*b −.016 .084 .039 .071 Step 4 .047* .001 .000 .000 .001 .003 .001 .003*  Practice .172 .132 .098 .111 .048 .179 .005  Child SEC .115 −.052 −.054 −.044 .054 .038 −.018 .240*  YA SEC .147 .097 −.054 −.107 .112 .213† .143 .031  MA SEC −.079 .111 .261*b .267*b −.014 .083 .041 .085  Current SES .219*a,b .037 −.007 −.014 .031 −.005 .028 .192* Total R2 .085* .056 .079 .070 .026 .077 .052 .103* n 112 112 110 110 110 110 110 110 Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027 .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a Step 2 .005 .022 .010 .003 .009 .062* .024 .003  Practice .157 .111 .079 .106 .043 .170 −.040  Child SEC .139 −.085 −.132 −.124 .059 .013 −.029 .221*  YA SEC .074 .160 .106 .058 .099 .265*a .164 .056 Step 3 .006 .008 .045* .047* .000 .005 .001 .003  Practice .165 .133 .101 .105 .049 .173 −.035  Child SEC .111 −.054 −.054 −.044 .054 .038 −.018 .242*  YA SEC .133 .095 −.054 −.106 .109 .214† .141 .012  MA SEC −.096 .107 .262*b .268*b −.016 .084 .039 .071 Step 4 .047* .001 .000 .000 .001 .003 .001 .003*  Practice .172 .132 .098 .111 .048 .179 .005  Child SEC .115 −.052 −.054 −.044 .054 .038 −.018 .240*  YA SEC .147 .097 −.054 −.107 .112 .213† .143 .031  MA SEC −.079 .111 .261*b .267*b −.014 .083 .041 .085  Current SES .219*a,b .037 −.007 −.014 .031 −.005 .028 .192* Total R2 .085* .056 .079 .070 .026 .077 .052 .103* n 112 112 110 110 110 110 110 110 Note: FDR = false discovery rate; MA = middle adult; SEC = socioeconomic context; SES = socioeconomic status; YA = young adulthood. aStatistically significant after FDR correction applied within each hierarchical step. bStatistically significant after FDR correction applied within each predictor. †p < .10. *p < .05. View Large Table 3. Hierarchical Regression Models of Child, Young Adult (YA), and Middle Adult (MA) SEC on Later-Life Cognition Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027 .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a Step 2 .005 .022 .010 .003 .009 .062* .024 .003  Practice .157 .111 .079 .106 .043 .170 −.040  Child SEC .139 −.085 −.132 −.124 .059 .013 −.029 .221*  YA SEC .074 .160 .106 .058 .099 .265*a .164 .056 Step 3 .006 .008 .045* .047* .000 .005 .001 .003  Practice .165 .133 .101 .105 .049 .173 −.035  Child SEC .111 −.054 −.054 −.044 .054 .038 −.018 .242*  YA SEC .133 .095 −.054 −.106 .109 .214† .141 .012  MA SEC −.096 .107 .262*b .268*b −.016 .084 .039 .071 Step 4 .047* .001 .000 .000 .001 .003 .001 .003*  Practice .172 .132 .098 .111 .048 .179 .005  Child SEC .115 −.052 −.054 −.044 .054 .038 −.018 .240*  YA SEC .147 .097 −.054 −.107 .112 .213† .143 .031  MA SEC −.079 .111 .261*b .267*b −.014 .083 .041 .085  Current SES .219*a,b .037 −.007 −.014 .031 −.005 .028 .192* Total R2 .085* .056 .079 .070 .026 .077 .052 .103* n 112 112 110 110 110 110 110 110 Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027 .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a Step 2 .005 .022 .010 .003 .009 .062* .024 .003  Practice .157 .111 .079 .106 .043 .170 −.040  Child SEC .139 −.085 −.132 −.124 .059 .013 −.029 .221*  YA SEC .074 .160 .106 .058 .099 .265*a .164 .056 Step 3 .006 .008 .045* .047* .000 .005 .001 .003  Practice .165 .133 .101 .105 .049 .173 −.035  Child SEC .111 −.054 −.054 −.044 .054 .038 −.018 .242*  YA SEC .133 .095 −.054 −.106 .109 .214† .141 .012  MA SEC −.096 .107 .262*b .268*b −.016 .084 .039 .071 Step 4 .047* .001 .000 .000 .001 .003 .001 .003*  Practice .172 .132 .098 .111 .048 .179 .005  Child SEC .115 −.052 −.054 −.044 .054 .038 −.018 .240*  YA SEC .147 .097 −.054 −.107 .112 .213† .143 .031  MA SEC −.079 .111 .261*b .267*b −.014 .083 .041 .085  Current SES .219*a,b .037 −.007 −.014 .031 −.005 .028 .192* Total R2 .085* .056 .079 .070 .026 .077 .052 .103* n 112 112 110 110 110 110 110 110 Note: FDR = false discovery rate; MA = middle adult; SEC = socioeconomic context; SES = socioeconomic status; YA = young adulthood. aStatistically significant after FDR correction applied within each hierarchical step. bStatistically significant after FDR correction applied within each predictor. †p < .10. *p < .05. View Large Post hoc sensitivity analyses further controlled for childhood SES proxies and evaluated associations between (a) SEC in childhood and adulthood on later-life cognition (Supplementary Table S1) and (b) SEC in childhood, young adulthood, and midlife on later-life cognition (Supplementary Table S2). The aforementioned significant effects remained, and no additional effects emerged. Residential Mobility and Cognition Last, we hypothesized that less residential mobility would be associated with better later-life cognition. Table 4 presents hierarchical regression results. Contrary to our hypothesis, significant effects were in the direction of higher residential mobility associating with better later-life cognition. Specifically, higher residential mobility in childhood (but not adulthood) was significantly associated with higher estimated crystallized intelligence (childhood: β = .194, p = .040; adulthood: β = .077, p = .42). Higher residential mobility in childhood tended to be associated with better immediate (β = .197, p = .040) and delayed (β = .185, p = .048) recall, and higher residential mobility in adulthood tended to be associated with better verbal learning (adulthood: β = .216, p = .029), however, these effects did not pass FDR corrections. Residential mobility in childhood and adulthood did not significantly relate to visuomotor speed, task switching speed or efficiency, or recognition memory. Table 4. Hierarchical Regression Models of Residential Mobility on Cognitive Functioning in Later Life Cognitive Construct Crystallized Intelligence Visuomotor Speed Task Switch Speed Task Switch Efficiency Verbal Learning Immediate Recall Delayed Recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .038* .023 .019 .013 .039 .040 .064* .015  Practice .150 .133 .108 .108 .023 .175 −.076  Moves in childhood .194*a,b −.030 −.033 −.043 .163† .197* .178† .099 Step 2 .006 .001 .019 .019 .042* .003 .029† .013  Practice .141 .089 .064 .171 .039 .228* −.041  Moves in childhood .198*a,b −.031 −.037 −.048 .171† .199* .185* .104  Moves in adulthood .077 −.029 −.145 −.146 .216* .054 .180† .121 Total R2 .043 .024 .038 .033 .082* .042 .093* .028 n 111 111 109 109 110 109 109 109 Cognitive Construct Crystallized Intelligence Visuomotor Speed Task Switch Speed Task Switch Efficiency Verbal Learning Immediate Recall Delayed Recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .038* .023 .019 .013 .039 .040 .064* .015  Practice .150 .133 .108 .108 .023 .175 −.076  Moves in childhood .194*a,b −.030 −.033 −.043 .163† .197* .178† .099 Step 2 .006 .001 .019 .019 .042* .003 .029† .013  Practice .141 .089 .064 .171 .039 .228* −.041  Moves in childhood .198*a,b −.031 −.037 −.048 .171† .199* .185* .104  Moves in adulthood .077 −.029 −.145 −.146 .216* .054 .180† .121 Total R2 .043 .024 .038 .033 .082* .042 .093* .028 n 111 111 109 109 110 109 109 109 Note: FDR = false discovery rate. aStatistically significant after FDR correction applied within each hierarchical step. bStatistically significant after FDR correction applied within each predictor. †p < .10. *p < .05. View Large Table 4. Hierarchical Regression Models of Residential Mobility on Cognitive Functioning in Later Life Cognitive Construct Crystallized Intelligence Visuomotor Speed Task Switch Speed Task Switch Efficiency Verbal Learning Immediate Recall Delayed Recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .038* .023 .019 .013 .039 .040 .064* .015  Practice .150 .133 .108 .108 .023 .175 −.076  Moves in childhood .194*a,b −.030 −.033 −.043 .163† .197* .178† .099 Step 2 .006 .001 .019 .019 .042* .003 .029† .013  Practice .141 .089 .064 .171 .039 .228* −.041  Moves in childhood .198*a,b −.031 −.037 −.048 .171† .199* .185* .104  Moves in adulthood .077 −.029 −.145 −.146 .216* .054 .180† .121 Total R2 .043 .024 .038 .033 .082* .042 .093* .028 n 111 111 109 109 110 109 109 109 Cognitive Construct Crystallized Intelligence Visuomotor Speed Task Switch Speed Task Switch Efficiency Verbal Learning Immediate Recall Delayed Recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .038* .023 .019 .013 .039 .040 .064* .015  Practice .150 .133 .108 .108 .023 .175 −.076  Moves in childhood .194*a,b −.030 −.033 −.043 .163† .197* .178† .099 Step 2 .006 .001 .019 .019 .042* .003 .029† .013  Practice .141 .089 .064 .171 .039 .228* −.041  Moves in childhood .198*a,b −.031 −.037 −.048 .171† .199* .185* .104  Moves in adulthood .077 −.029 −.145 −.146 .216* .054 .180† .121 Total R2 .043 .024 .038 .033 .082* .042 .093* .028 n 111 111 109 109 110 109 109 109 Note: FDR = false discovery rate. aStatistically significant after FDR correction applied within each hierarchical step. bStatistically significant after FDR correction applied within each predictor. †p < .10. *p < .05. View Large Discussion The present study utilized a novel method of measuring the relative wealth and quality of the environment, or SEC, to predict cognitive functioning among generally healthy older adults. Census data on county-level unemployment rate, median household income, and percent of individuals with a high school diploma were obtained from reported lifetime residential addresses to compute SEC. Our primary focus was on the influences of broad socioeconomic environmental conditions, particularly in early life, on later-life cognition. Higher childhood SEC (ages 0–18), controlling for practice effects, adulthood SEC, and current SES, was significantly associated with better recognition memory and tended to be associated with higher estimated crystallized intelligence. Notably, of the verbal memory functions assessed in the present study, recognition memory is most reflective of retention and least impacted by executive influence on memory retrieval (i.e., not reliant on free recall; Lezak, 2004). Interestingly, in contrast with childhood SEC, adulthood SEC effects tended to be executive. Higher adulthood SEC (ages 19–60), controlling for practice effects, childhood SEC, and current SES, was significantly associated with better immediate verbal memory recall, whereas higher midlife SEC (ages 40–60) was significantly associated with faster task switching and better task switching efficiency. Immediate verbal memory requires free recall, which has an executive retrieval component (Lezak, 2004). Therefore, in the broad community context, early socioeconomic contexts (childhood SEC) may impact the development of cognitive reserve and long-term integrity of verbal memory functioning, whereas more proximal socioeconomic environmental conditions (midlife SEC) may impact higher-level cognitive functions, which may be among the first to deteriorate with aging (Salthouse, Atkinson, & Berish, 2003). Together, these findings suggest there may be a quadratic function in the sensitive periods of SEC effects on later-life cognition. Importantly, these SEC findings were independent of current individual-level SES. Although the present investigation focused on broader environmental socioeconomic conditions, individual-level SES also affects later-life cognition. In line with previous evidence (Gottfredson, 2004), higher current SES (controlling for practice effects and SEC) was associated with higher estimated crystallized intelligence. This association likely reflects bidirectional relationships between crystallized intelligence with elements of SES including education and occupational status. Contrary to hypotheses, higher residential mobility in childhood was associated with better cognitive function in later-life, specifically higher estimated crystallized intelligence. For some cohorts, life stages, and subcultural contexts, an individual’s or family’s ability to move residences may reflect intellectual and socioeconomic resources that result in the freedom to change employment or residence as opposed to economic instability. In the present study, higher residential mobility in adulthood was associated with higher current SES. Our study contributes a new methodological approach to measure SEC. Advantages of this methodology are that objective address data are less likely to be misremembered (Ward, 2011) and unlikely to be influenced by a positivity bias in older adults (Carstensen, 2006). Conversely, subjective reports of socioeconomic position and adversity in childhood can be misremembered in a positive light, diminishing the accuracy of these estimates (Maughan & Rutter, 1997). Another advantage of this SEC approach is that multiple parameters can be derived from these data. In the present study, total SEC exposure in childhood and adulthood, as well as residential mobility, were calculated. The high stability of SEC before age 19 in the present sample made smaller distinctions (e.g., early vs middle childhood vs adolescence) unfeasible; however, we explored additional developmental periods within adulthood (i.e., young vs middle adulthood). It should be noted that SEC is particularly refined in the present sample with regard to individual exposures because the majority of lifetime residential addresses were in Kentucky. Kentucky has the fourth highest number of counties in the United States, with 120 counties each covering only an average of 337 square miles. Although there was sufficient SEC variability, it is important to examine this method in other geographic locations in the future. Additionally, the present study employed county-level census data to capture the broad environmental context. Future investigations might explore using other census measures such as census tracts (which were not uniformly collected across the United States until the 2000 Census and therefore not used in the present study) to capture the nuances of SEC exposure. An important consideration in the present study is whether SEC contributes new information to our understanding of social stratification and health above and beyond individual-level measures of SES. Although additional research is needed to empirically test this question, our results suggest that SEC and SES are nonoverlapping constructs that may provide unique information (e.g., correlations between SES and SEC were small and not significant; Table 1). Additionally, SEC effects on later life cognition were independent of current and childhood individual-level SES. SEC may reflect health-relevant exosystem-level factors, such as physical environmental exposures and access to health care, that affect later life cognition (e.g., Cherrie et al., 2018). To test this interpretation of the representativeness of our SEC variable, we conducted a post hoc analysis using county health rankings from the University of Wisconsin (UW) Population Health Institute (Remington, Catlin, & Gennuso, 2015). The UW data contain annual (2010–2017) county-level measures of “health by place” within two main domains: health outcomes (HO), which include quality of life, morbidity, and mortality data; and health factors (HF), which include socioeconomic factors (education, employment, income, social support, community safety/crime), access to and quality of healthcare, health behaviors (e.g., tobacco and drug use), and physical environmental factors (air and water quality, housing, and transportation). Lower county-level HO and HF values represent higher within-state rankings. We identified lifelong Kentucky residents in our data whose lifetime SEC scores (averaged across ages 0–60) were within the top and bottom 10% of the SEC distribution (n = 9 individuals representing 20 counties). For each of the nine participants, we averaged the HO and HF rankings across all counties s/he lived in. In post hoc analyses, higher SEC was significantly associated with better HO rankings (r = −.82, p = .007) and better HF rankings (r = −.94, p < .001). Taken together, our results suggest that SES and SEC represent distinct constructs and that our SEC measure may capture a range of exosystem-level variables that operate above individual-level SES. Although there was a range in the indicators of current SES, the present sample was majority Caucasian, relatively wealthy, and well educated. The sample also demonstrated above average estimated premorbid intelligence (NAART FSIQ M = 112.45, range = 87.24–127.02). These characteristics may limit the generalizability of the current findings. Lifespan SEC trajectories were equifinal in that all participants were living in a relatively affluent county at assessment (Figure 1). Insofar as high SES and concurrent high SEC protect against adverse effects of early experience on cognition (Wight et al., 2006), results from this sample may reflect only the influence of childhood SEC that was resistant to these buffers. Additionally, although the effects of childhood SEC on later life cognition were independent of childhood SES (assessed using two items from the CTQ that broadly indicate material deprivation and may inexactly measure childhood SES), future research should corroborate this using other indicators of childhood SES (e.g., parental education, employment, and/or income). Furthermore, future work is needed that simultaneously examines whether and how microsystem-level variables (e.g., individual-level SES and childhood enrichment factors, such as number of children’s books in the home or frequency of someone in the home reading aloud to the child, etc.) and exosystem-level variables (i.e., SEC, green spaces, etc.) individually and interactively influence later-life cognition. Regarding the feasibility of the SEC approach, there was a low percentage of missing SEC data in the present sample. However, it should be noted that participants may misremember the exact years they lived at a certain address. Finally, the present investigation represents a preliminary examination of SEC on later life cognition; although all reported analyses correspond to theoretically and empirically informed a priori hypotheses, multiple analyses were conducted. Therefore, these effects require replication in large, heterogeneous samples. SEC may be a useful adjunct to conventional measures of early life SES, which may be prone to biases when self-reported. The SEC measure may benefit from further refinement in terms of indicators and categorization of SEC periods (childhood vs adulthood). The present study focused on SEC total exposure and residential mobility, but other parameters could be derived (e.g., SEC range or individual standard deviation as reflections of variability in SEC environments). Additionally, separate components of SEC could be investigated if theorized to affect later life cognition in unique ways. For example, particular aspects of SEC (e.g., school zone ratings, proximity to libraries) during certain developmental periods (e.g., early childhood) may be more strongly associated with cognitive reserve development. Further research can more fully establish the validity of this approach and the relationship between SEC and later-life health and cognitive outcomes. In addition, how SEC gets “under the skull” to affect cognitive development is an important question. Some pathways may be relatively direct and only involve the individual and the exosystem (e.g., exposure to environmental toxins and pollutants), but others may involve interactions among ecological levels, such as interactions between exosystem employment levels and microsystem family environments (Bronfenbrenner, 1977). Mediational models may also be fruitful; for example, employment levels might impact individual cognitive development through school quality. Socioeconomic-related health disparities exist for cognitive decline, dementia, and Alzheimer’s disease (D. A. Evans et al., 1997; Yaffe et al., 2013; Zhang et al., 2016). Understanding the risk factors for these diseases in the context of broader environmental influences is a critical first step to developing effective ways to promote healthy cognitive aging. Measuring SEC opens the door for an ecological understanding of lifespan cognitive development, with implications for later-life cognitive function. Supplementary Material Supplementary data are available at The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences online. Funding This work was supported by the Dana Foundation and the National Institutes of Health (R01-AG026307, K02-033629, K99-AG056635, P30-AG028383, and CTSA-UL1TR000117). Conflict of Interest The authors report no conflicts of interest. Acknowledgments Author Contributions: A. B. Scott and R. G. Reed contributed equally to this work. S. C. Segerstrom designed the parent study, including the measures. A. B. Scott and S. C. Segerstrom planned the study, and with R. G. Reed, performed the data analyses and wrote the article. N. E. Garcia-Willingham provided consultation for the neuropsychological measures and contributed to writing the article. K. A. Lawrence contributed to data analyses and writing the article. References American Fact Finder . ( 2017 ). Annual estimates of the resident population for selected age groups by sex . Retrieved November 5, 2017 from https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?src=CF Barnes , L. L. , Mendes de Leon , C. F. , Wilson , R. S. , Bienias , J. L. , & Evans , D. A . ( 2004 ). Social resources and cognitive decline in a population of older African Americans and whites . Neurology , 63 , 2322 – 2326 . doi: https://doi.org/10.1212/01.wnl.0000147473.04043.b3 Benjamini , Y. , & Hochberg , Y . ( 1995 ). Controlling the false discovery rate: A practical and powerful approach to multiple testing . Journal of the Royal Statistical Society Series B (Methodological) , 57 , 289 – 300 . doi:10.2307/2346101 Berkman , L. F. , Seeman , T. E. , Albert , M. , Blazer , D. , Kahn , R. , Mohs , R. ,… Rowe , J . ( 1993 ). High, usual and impaired functioning in community-dwelling older men and women: Findings from the MacArthur Foundation Research Network on Successful Aging . Journal of Clinical Epidemiology , 46 , 1129 – 1140 . doi: https://doi.org/10.1016/0895-4356(93)90112-E Bernstein , D. P. , Stein , J. A. , Newcomb , M. D. , Walker , E. , Pogge , D. , Ahluvalia , T. ,… Zule , W . ( 2003 ). Development and validation of a brief screening version of the Childhood Trauma Questionnaire . Child Abuse & Neglect , 27 , 169 – 190 . doi: https://doi.org/10.1016/S0145-2134(02)00541-0 Blair , J. R. , & Spreen , O . ( 1989 ). Predicting premorbid IQ: A revision of the national adult reading test . Clinical Neuropsychologist , 3 , 129 – 136 . doi: https://doi.org/10.1080/13854048908403285 Bronfenbrenner , U . ( 1977 ). Toward an experimental ecology of human development . American Psychologist , 32 , 513 – 531 . doi: https://doi.org/10.1037//0003-066x.32.7.513 Carstensen , L. L . ( 2006 ). The influence of a sense of time on human development . Science , 312 , 1913 – 1915 . doi: https://doi.org/10.1126/science.1127488 Cassarino , M. , O’Sullivan , V. , Kenny , R. A. , & Setti , A . ( 2016 ). Environment and cognitive aging: A cross-sectional study of place of residence and cognitive performance in the Irish longitudinal study on aging . Neuropsychology , 30 , 543 – 557 . doi: https://doi.org/10.1037/neu0000253 Charles , S. T. , Mather , M. , & Carstensen , L. L . ( 2003 ). Aging and emotional memory: The forgettable nature of negative images for older adults . Journal of Experimental Psychology , 132 , 310 – 324 . doi: https://doi.org/10.1037/0096-3445.132.2.310 Cherrie , M. P. , Shortt , N. K. , Mitchell , R. J. , Taylor , A. M. , Redmond , P. , Thompson , C. W. ,… Pearce , J. R . ( 2018 ). Green space and cognitive ageing: A retrospective life course analysis in the Lothian Birth Cohort 1936 . Social Science & Medicine (1982) , 196 , 56 – 65 . doi: https://doi.org/10.1016/j.socscimed.2017.10.038 Christensen , H . ( 2001 ). What cognitive changes can be expected with normal ageing ? The Australian and New Zealand Journal of Psychiatry , 35 , 768 – 775 . doi: https://doi.org/10.1046/j.1440-1614.2001.00966.x Drane , D. L. , Yuspeh , R. L. , Huthwaite , J. S. , & Klingler , L. K . ( 2002 ). Demographic characteristics and normative observations for derived-trail making test indices . Neuropsychiatry, Neuropsychology, and Behavioral Neurology , 15 , 39 – 43 . Duncan , G. J. , Brooks-Gunn , J. , & Klebanov , P. K . ( 1994 ). Emotion deprivation and early childhood development . Child Development , 65 , 296 – 318 . doi: https://doi.org/10.2307/1131385 Evans , D. A. , Hebert , L. E. , Beckett , L. A. , Scherr , P. A. , Albert , M. S. ,… & Taylor , J. O . ( 1997 ). Education and other measures of socioeconomic status and risk of incident Alzheimer disease in a defined population of older persons . Archives of Neurology , 54 , 1399 – 1405 . doi: https://doi.org/10.1001/archneur.1997.00550230066019 Evans , G. W. , & Kantrowitz , E . ( 2002 ). Socioeconomic status and health: The potential role of environmental risk exposure . Annual Review of Public Health , 23 , 303 – 331 . doi: https://doi.org/10.1146/annurev.publhealth.23.112001.112349 Fotenos , A. F. , Mintun , M. A. , Snyder , A. Z. , Morris , J. C. , & Buckner , R. L . ( 2008 ). Brain volume decline in aging: Evidence for a relation between socioeconomic status, preclinical Alzheimer disease, and reserve . Archives of Neurology , 65 , 113 – 120 . doi: https://doi.org/10.1001/archneurol.2007.27 Gottfredson , L. S . ( 2004 ). Intelligence: Is it the epidemiologists’ elusive “fundamental cause” of social class inequalities in health ? Journal of Personality and Social Psychology , 86 , 174 – 199 . doi: https://doi.org/10.1037/0022-3514.86.1.174 Hackman , D. A. , Farah , M. J. , & Meaney , M. J . ( 2010 ). Socioeconomic status and the brain: Mechanistic insights from human and animal research . Nature Reviews. Neuroscience , 11 , 651 – 659 . doi: https://doi.org/10.1038/nrn2897 Hollingshead , A. B . ( 1975 ). Four factor index of social status . New Haven, CT : Yale University . Unpublished manuscript. Ivnik , R. J. , Malec , J. F. , Tangalos , E. G. , Petersen , R. C. , Kokmen , E. , & Kurland , L. T . ( 1990 ). The Auditory-Verbal Learning Test (AVLT): Norms for ages 55 years and older . Psychological Assessment , 2 , 304 – 312 . doi: https://doi.org/10.1037//1040-3590.2.3.304 Jelleyman , T. , & Spencer , N . ( 2007 ). Residential mobility in childhood and health outcomes: A systematic review . Journal of Epidemiology and Community Health , 62 , 584 – 592 . doi: https://doi.org/10.1136/jech.2007.060103 Jiang , T. , Yu , J. T. , Tian , Y. , & Tan , L . ( 2013 ). Epidemiology and etiology of Alzheimer’s disease: From genetic to non-genetic factors . Current Alzheimer Research , 10 , 852 – 867 . doi: https://doi.org/10.2174/15672050113109990155 Kaplan , G. A. , Turrell , G. , Lynch , J. W. , Everson , S. A. , Helkala , E. L. , & Salonen , J. T . ( 2001 ). Childhood socioeconomic position and cognitive function in adulthood . International Journal of Epidemiology , 30 , 256 – 263 . doi: https://doi.org/10.1093/ije/30.2.256 Kauhanen , L. , Lakka , H. M. , Lynch , J. W. , & Kauhanen , J. ( 2006 ). Social disadvantages in childhood and risk of all-cause death and cardiovascular disease in later life: A comparison of historical and retrospective childhood information. International Journal of Epidemiology, 35, 962–968. doi: https://doi.org/10.1093/ije/dyl046 Krieger , N . ( 2001 ). A glossary for social epidemiology . Journal of Epidemiology & Community Health , 55 , 693 – 700 . doi: https://doi.org/10.1136/jech.55.10.693 Krieger , N. , Rehkopf , D. H. , Chen , J. T. , Waterman , P. D. , Marcelli , E. , & Kennedy , M . ( 2008 ). The fall and rise of US inequities in premature mortality: 1960–2002 . PLoS Medicine , 5 , e46 . doi: https://doi.org/10.1371/journal.pmed.0050046 Krieger , N. , Williams , D. R. , & Moss , N. E . ( 1997 ). Measuring social class in US public health research: Concepts, methodologies, and guidelines . Annual Review of Public Health , 18 , 341 – 378 . doi: https://doi.org/10.1146/annurev.publhealth.18.1.341 Lanphear , B. P . ( 2015 ). The impact of toxins on the developing brain . Annual Review of Public Health , 36 , 211 – 230 . doi: https://doi.org/10.1146/annurev-publhealth-031912-114413 Lezak , M. D . ( 2004 ). Neuropsychological assessment . New York, NY: Oxford University Press . Luo , Y. , & Waite , L. J . ( 2005 ). The impact of childhood and adult SES on physical, mental, and cognitive well-being in later life . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 60 , 93 – 101 . doi: https://doi.org/10.1093/geronb/60.2.s93 Lupien , S. J. , McEwen , B. S. , Gunnar , M. R. , & Heim , C . ( 2009 ). Effects of stress throughout the lifespan on the brain, behaviour and cognition . Nature Reviews. Neuroscience , 10 , 434 – 445 . doi: https://doi.org/10.1038/nrn2639 Maughan , B. , & Rutter , M . ( 1997 ). Retrospective reporting of childhood adversity: Issues in assessing long-term recall . Journal of Personality Disorders , 11 , 19 – 33 . doi:10.1521/pedi.1997.11.1.19 McLeod , J. D. , & Kessler , R. C . ( 1990 ). Socioeconomic status differences in vulnerability to undesirable life events . Journal of Health and Social Behavior , 31 , 162 – 172 . doi: https://doi.org/10.2307/2137170 Miller , G. E. , Chen , E. , & Parker , K. J . ( 2011 ). Psychological stress in childhood and susceptibility to the chronic diseases of aging: Moving toward a model of behavioral and biological mechanisms . Psychological Bulletin , 137 , 959 – 997 . doi: https://doi.org/10.1037/a0024768 Reitan , R. M . ( 1955 ). The relationship of the Trail Making Test to organic brain damage . Journal of Consulting Psychology , 19 , 393 – 394 . doi: https://doi.org/10.1037/h0044509 Remington , P. L. , Catlin , B. B. , & Gennuso , K. P . ( 2015 ). The county health rankings: Rationale and methods . Population Health Metrics , 13 , 11 . doi: https://doi.org/10.1186/s12963-015-0044-2 Salthouse , T. A. , Atkinson , T. M. , & Berish , D. E . ( 2003 ). Executive functioning as a potential mediator of age-related cognitive decline in normal adults . Journal of Experimental Psychology. General , 132 , 566 – 594 . doi: https://doi.org/10.1037/0096-3445.132.4.566 Santos , D. N. , Assis , A. M. , Bastos , A. C. , Santos , L. M. , Santos , C. A. , Strina , A. ,… Barreto , M. L . ( 2008 ). Determinants of cognitive function in childhood: A cohort study in a middle income context . BMC Public Health , 8 , 202 . doi: https://doi.org/10.1186/1471-2458-8-202 Sattler , C. , Toro , P. , Schönknecht , P. , & Schröder , J . ( 2012 ). Cognitive activity, education and socioeconomic status as preventive factors for mild cognitive impairment and Alzheimer’s disease . Psychiatry Research , 196 , 90 – 95 . doi: https://doi.org/10.1016/j.psychres.2011.11.012 Straughen , J. K. , Caldwell , C. H. , Osypuk , T. L. , Helmkamp , L. , & Misra , D. P . ( 2013 ). Direct and proxy recall of childhood socio-economic position and health . Paediatric and Perinatal Epidemiology , 27 , 294 – 302 . doi: https://doi.org/10.1111/ppe.12045 Strauss , E. , Sherman , E. M. , & Spreen , O . ( 2006 ). A compendium of neuropsychological tests: Administration, norms, and commentary . New York : Oxford University Press . Thomas , T. L. , DiClemente , R. , & Snell , S . ( 2014 ). Overcoming the triad of rural health disparities: How local culture, lack of economic opportunity, and geographic location instigate health disparities . Health Education Journal , 73 , 285 – 294 . doi: https://doi.org/10.1177/0017896912471049 Ward , M. M . ( 2011 ). Concordance of sibling’s recall of measures of childhood socioeconomic position . BMC Medical Research Methodology , 11 , 147 . doi: https://doi.org/10.1186/1471-2288-11-147 Wight , R. G. , Aneshensel , C. S. , Miller-Martinez , D. , Botticello , A. L. , Cummings , J. R. , Karlamangla , A. S. , & Seeman , T. E . ( 2006 ). Urban neighborhood context, educational attainment, and cognitive function among older adults . American Journal of Epidemiology , 163 , 1071 – 1078 . doi: https://doi.org/10.1093/aje/kwj176 Wright , C. B. , Sacco , R. L. , Rundek , T. , Delman , J. , Rabbani , L. , & Elkind , M . ( 2006 ). Interleukin-6 is associated with cognitive function: The Northern Manhattan Study . Journal of Stroke and Cerebrovascular Diseases , 15 , 34 – 38 . doi: https://doi.org/10.1016/j.jstrokecerebrovasdis.2005.08.009 Yaffe , K. , Falvey , C. , Harris , T. B. , Newman , A. , Satterfield , S. , Koster , A. ,… Simonsick , E .; Health ABC Study . ( 2013 ). Effect of socioeconomic disparities on incidence of dementia among biracial older adults: Prospective study . BMJ (Clinical research ed.) , 347 , f7051 . doi:10.1136/bmj.f7051 Yang , Y. C. , Gerken , K. , Schorpp , K. , Boen , C. , & Harris , K. M . ( 2017 ). Early-life socioeconomic status and adult physiological functioning: A life course examination of biosocial mechanisms . Biodemography and Social Biology , 63 , 87 – 103 . doi: https://doi.org/10.1080/19485565.2017.1279536 Zhang , Z. , Hayward , M. D. , & Yu , Y. L . ( 2016 ). Life course pathways to racial disparities in cognitive impairment among older Americans . Journal of Health and Social Behavior , 57 , 184 – 199 . doi: https://doi.org/10.1177/0022146516645925 © The Author(s) 2018. Published by Oxford University Press on behalf of The Gerontological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The Journals of Gerontology Series B: Psychological Sciences and Social Sciences Oxford University Press

Loading next page...
 
/lp/ou_press/lifespan-socioeconomic-context-associations-with-cognitive-functioning-zc5OX5AlGY
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of The Gerontological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
1079-5014
eISSN
1758-5368
D.O.I.
10.1093/geronb/gby071
Publisher site
See Article on Publisher Site

Abstract

Abstract Objectives Early socioeconomic status (SES) correlates with later-life cognition. However, the effect of socioeconomic context (SEC), which reflects influences from broader ecological contexts, has not been examined. The present study developed a measure of SEC using lifetime residential addresses and examined SEC and residential mobility effects on later-life cognition. Method Older adults (N = 117, Mage = 75) reported addresses since birth. Latent SEC was constructed from census income, employment, and education (1920–2010) for each county and census year, extrapolated between census years. Controlling for current SES, SEC in childhood (ages 0–18) and adulthood (ages 19–60), with finer granulations in young adulthood (ages 19–39) and midlife (ages 40–60), predicted later-life cognition. Effects of residential mobility on later-life cognition were also examined. Results Higher childhood and adulthood SEC were associated with better Auditory Verbal Learning Test recognition (β = .24, p = .012) and immediate recall (β = .26, p = .008). Higher midlife SEC was associated with faster task switching (β = .26, p = .025) and better task switching efficiency (β = .27, p = .022). Higher residential mobility in childhood was associated with higher crystallized intelligence (β = .194, p = .040). Discussion Independent of current SES, childhood and adulthood SEC predicted later-life cognition, which may be sensitive to effects of social institutions and environmental health. SEC assessed across the lifespan, and related residential mobility information may be important complements to SES in predicting later-life cognitive health. Development, Early life, Executive functions, IQ, Memory, Socioeconomic One of the best-studied individual influences on the degree and rate of cognitive development is socioeconomic status (SES), defined as an individual’s social position in relation to others and often operationalized as some combination of income, education, and occupation (Krieger, 2001; Krieger, Williams, & Moss, 1997). Development, however, occurs in context. Ecological systems theory (Bronfenbrenner, 1977) embeds the individual in a close social context (microsystem), a broader community context (exosystem), and an even broader cultural context (macrosystem), all of which interact with each other and all of which can influence development and health, including cognitive health. Socioeconomic context (SEC) is an exosystem-level variable that reflects the relative wealth and quality of an individual’s community, which in turn correlates with factors such as environmental exposures, educational quality, and access to health care. SEC and SES can be operationalized using similar variables at different levels of analysis (e.g., county- vs individual-level income, education, and unemployment rate or occupation information). However, SEC is not simply an expanded measure of SES but captures emergent properties at broader contextual levels (not always reflected at the individual level) that arise from interactions among historical, environmental, and community circumstances (G. W. Evans & Kantrowitz, 2002; Krieger et al., 2008; McLeod & Kessler, 1990; Yang, Gerken, Schorpp, Boen, & Harris, 2017). For example, in rural communities, local culture, economy, and geographic location determine health disparities via local tax base size, poverty rates, and access to education and health care (Thomas, DiClemente, & Snell, 2014). The importance of SEC is emphasized by Bronfenbrenner’s (1977) proposition that “research on the ecology of human development requires investigations that go beyond the immediate setting containing the person to examine the larger contexts, both formal and informal, that affect events within the immediate setting” (p. 527). The overlap between SES and SEC may explain the relationship between family income and environmental exposures including air pollutants, water quality, and ambient noise, as well as the quality of residences, educational facilities, and neighborhoods (G. W. Evans & Kantrowitz, 2002). However, SEC effects may trump SES effects; in the United States, the increased risk for premature mortality associated with low SES is ameliorated for people who lived in high SEC environments (Krieger et al., 2008). Thus, SEC captures resources and exposures that are not necessarily reflected at the individual level. SEC may affect cognitive development and result in differences in neuropsychological performance in older age. Cognitive and brain development may be negatively affected by some environmental exposures associated with SEC such as environmental toxins and pollutants (Lanphear, 2015). In contrast, exposure to green space (e.g., public parks) during childhood and adulthood may support healthy cognitive aging in later life (Cherrie et al., 2018). Other exposures may affect the brain because they are stressful (e.g., noise, crowding). Stress is associated with higher systemic inflammation, which, in turn, may compromise cognition and brain health, particularly in early life (Lupien, McEwen, Gunnar, & Heim, 2009; Miller, Chen, & Parker, 2011; Wright et al., 2006). At the same time, residing in potentially more stimulating urban (vs rural) environments, particularly during childhood, is associated with higher global and executive cognitive function (Cassarino, O’Sullivan, Kenny, & Setti, 2016). Childhood may be a critical period for SEC exposure. Less socioeconomic hardship in this developmental stage may increase the probability of successful aging, defined as being relatively free from health and psychological problems in older age (Berkman et al., 1993). Better home and neighborhood environmental conditions (e.g., sanitary infrastructure) and greater concentrations of affluent neighbors are associated with better cognition in children (Duncan, Brooks-Gunn, & Klebanov, 1994; Santos et al., 2008). These early life cognitive differences may set people on certain trajectories that carry into older age. Childhood SEC may also provide certain protective resources, over and above SES, on later-life cognition. For example, among older adults who report lower childhood SES, those that live in relatively affluent urban neighborhoods fare better cognitively than their counterparts in more disadvantaged urban neighborhoods (Wight et al., 2006). Although the current study focuses on SEC, we recognize that individual-level SES is an important determinant of adult cognition. Cross-sectionally and longitudinally, lower SES is associated with worse cognition and higher risk of cognitive aging disorders (Barnes, Mendes de Leon, Wilson, Bienias, & Evans, 2004; Fotenos, Mintun, Snyder, Morris, & Buckner, 2008), whereas higher SES is protective and associated with lower risk of cognitive decline, including Alzheimer’s disease (D. A. Evans et al., 1997; Hackman, Farah, & Meaney, 2010; Jiang, Yu, Tian, & Tan, 2013; Sattler, Toro, Schönknecht, & Schröder, 2012). Additionally, higher SES in childhood is associated with better cognition function in older age (Kaplan et al., 2001; Luo & White, 2005). This investigation is the first to empirically test whether childhood and adulthood SEC contribute to later-life cognition above and beyond current individual-level SES and, in additional post hoc sensitivity analyses, above and beyond broad proxies of childhood SES. Assessment of Lifetime SEC A challenge in assessing lifetime socioeconomic conditions among older adults, whether individual status or community context, is reliance on retrospective self-reports. First, older adults must remember increasingly temporally distant information to provide a picture of earlier SES or SEC, potentially resulting in less accurate reports. Second, older adults typically have positive cognitive biases that may affect the accuracy of their memories of people and situations (Charles, Mather, & Carstensen, 2003). Specifically, in the positivity effect, focus shifts from negative information in youth to the reciprocal with age such that older adults’ memories are more likely biased toward positive information (Carstensen, 2006). For example, compared with objective data, older adults self-reported less childhood adversity (Maughan & Rutter, 1997), suggesting that they may be more forgetful of negative autobiographical events or have a positively biased memory for childhood experiences or adverse conditions. Inaccuracies in older adults’ recollections of subjective socioeconomic proxies can compromise attempts to link environments in childhood to health in older age. As an illustration, older men’s self-reported recall of childhood SES was not associated with premature mortality or coronary risk, but social disadvantage assessed from objective data including school health records was robustly associated with higher all-cause mortality risk (odds ratio [OR] = 1.41) and acute coronary event risk (OR = 1.50; Kauhanen et al., 2006). In the present study, older adults reported their lifetime residential addresses in an attempt to generate a less subjective and potentially more accurate measure of lifetime SEC. Objective information such as addresses are less subject to positive memory biases and may be recalled with fewer errors than subjective measures such as individual perceptions of socioeconomic proxies (Straughen, Caldwell, Osypuk, Helmkamp, & Misra, 2013; Ward, 2011). By linking these addresses to county-level data from the U.S. Census, participants’ residences were linked to the broad SEC of each county at particular points in time. Reports of lifetime residential addresses also permitted examination of residential mobility, the frequency with which an individual changes residential location in a specified time period. The frequency of moves may be especially helpful in investigating a dose–response effect of stressors associated with residential change. Higher residential mobility in childhood and adulthood is associated with poverty, unemployment, and family disruption (Jelleyman & Spencer, 2007). The Present Study The present investigation represents the first use of lifetime residential addresses to investigate effects of lifetime SEC and residential mobility on later-life cognition. We hypothesized that higher SEC, particularly in early life, would be associated with better later-life cognition, above and beyond current individual-level SES. We also hypothesized that less residential mobility would be associated with the better cognitive function. To provide a broad assessment of cognition in domains that may be vulnerable to socioeconomic influences, several neuropsychological functions were assessed, including estimated crystallized intelligence, visuomotor speed and task switching speed and efficiency, and verbal memory (verbal learning, immediate recall, delayed recall, and recognition). Most of the cohort (97%) lived in the same county at the time of neuropsychological testing, with 3% living in adjacent communities, providing equifinality with regard to SEC. Therefore, the present study tested the effects of lifetime county-level SEC on later-life cognition where current county-level SEC was constant across the cohort. Finally, to explore potential SEC effects specific to developmental periods, we tested associations between childhood SEC, young adulthood SEC, and midlife SEC on later-life cognition. Method Participants Participants were 117 community-dwelling older adults over the age of 60 (Mage= 75 years, range = 60–92) participating in an ongoing longitudinal study of psychological, cognitive, and immunological health. The sample was 60% female, consistent with the general older adult population according to the 2010 U.S. Census (57% female aged 65 years and older; American Fact Finder, 2017). A majority (93%) of the sample was Caucasian, with the remainder African American. Median education was 16 years (range = 9–22 years), and median household income was $78,500 (range = $9,000–$400,000). Because the study involved measurement of immune parameters, exclusion criteria were: diseases or disorders affecting the immune system; radiation therapy or chemotherapy within 5 years; unwillingness to undergo venipuncture; or more than two medications in the classes of psychotropics, antihypertensives, hormone replacement, or thyroid supplements. Additionally, use of any of the following medications was exclusionary: medications for cognitive impairment (e.g., donepezil, memantine, rivastigmine), opiates, systemic steroids, cytotoxic drugs, and tumor necrosis factor blockers. Thus, participants represent a generally healthy community sample of older adults. The majority of the sample scored within normal limits on the neuropsychological measures. Eight individuals scored in the clinical range (Z < −2.0) on 1–3 parameters of the Auditory Verbal Learning Test (AVLT), but their scores on the other AVLT parameters and the Trail Making Test were in the normal range. Design and Procedures Participants were recruited from a prospective research pool maintained by the Sanders-Brown Center on Aging at the University of Kentucky. In the longitudinal study, participants are interviewed, undergo neuropsychological testing, and have blood drawn every 6 months. The present study utilized only the neuropsychological assessments performed at the initial assessment, when participants provided lifetime residential addresses, to reduce the influence of practice effects. This study was conducted with the approval of the University of Kentucky Institutional Review Board, and all participants provided informed consent for study procedures. Measures Socioeconomic context SEC was measured using self-reported addresses from birth until age 60. Addresses were coded for county-level information using the publicly accessible database Social Explorer (https://www.socialexplorer.com), which provides access to U.S. Census Bureau data from 1790 to 2010. In the event a participant could not remember an address, they were asked if they could provide a specific location or landmark near that residence (e.g., township name, cross streets, nearby hospital or school) to enable researchers to identify the county of residence. Data were missing (details below) if participants could not recall either an address or usable landmark for a particular residence. For each county of residence reported in the sample, the percent unemployed, percent with at least a high school diploma, and/or median household income (in 2010 dollars) were obtained for each census year from 1920 to 2010. The percent unemployed was available for decades 1930–1950 and 1970–2010; percent with at least a high school diploma, 1920–1950 and 1970–2010; and median household income, 1970–2010. Data for the 1960 Census were not yet analyzed by the U.S. Census Bureau and therefore not available. Values for percent unemployed and percent with at least a high school diploma were squared to normalize their distributions. Variables were standardized (M = 0, SD = 1) within census years to remove between-census variation (i.e., progressive increases in education levels over the century) while maintaining between-county relationships. Education, income, and employment values from all counties and census years were used to construct an SEC latent variable in Mplus; scores were output for each county and census year. Noncensus year scores were generated by extrapolating between census years for each county. The chi-square test was statistically significant, as is often true with large sample sizes [χ2(3) = 985.44, p < .001], but fit indices indicated that the model fits well (comparative fit index = 1.00; Tucker–Lewis index = 1.00; AIC = 14771.3; BIC = 14822.6). All three indicators had statistically significant loadings on the SEC latent variable (education, β = .86, p < .001; income, β = .78, p < .001; employment, β = .30, p < .001). To account for missing SEC data (5.4% person-year observations) due to the overseas residence and active military duty (1.9%) and failure to recall a residence (3.5%), we implemented multiple imputation using the Amelia package (version 1.7.3) in R (version 3.0.3). This method implements an expectation–maximization with bootstrapping algorithm to impute missing values. The variables used to generate imputed scores were person, year, SEC, and county. A model including a linear effect of time was used to generate imputed data sets. SEC scores from five imputed data sets were averaged together to obtain the final SEC scores. Participants were assigned the relevant SEC value for each year of their life from birth until age 60 based on the county in which they lived at each age. Figure 1 illustrates SEC data through age 60 for five selected participants. Childhood SEC was calculated by averaging the SEC values for each participant from birth through age 18. Adulthood SEC was calculated by averaging the SEC values for ages 19 through 60. In addition, for exploratory analyses, adulthood SEC was subdivided into young adulthood SEC (ages 19–39) and midlife SEC (ages 40–60), which were calculated in the same manner. Figure 1. View largeDownload slide Example socioeconomic context (SEC) trajectories. This figure illustrates SEC scores for five representative individuals through age 60. Higher SEC values correspond to higher wealth and quality of an individual’s community. Figure 1. View largeDownload slide Example socioeconomic context (SEC) trajectories. This figure illustrates SEC scores for five representative individuals through age 60. Higher SEC values correspond to higher wealth and quality of an individual’s community. Residential mobility Using the lifetime address reports, residential mobility was calculated as the number of moves a participant made, including moves within the same county and to a different county. Residential mobility was calculated for childhood (ages 0–18) and adulthood (ages 19–60). Socioeconomic status To control for the effect of current individual SES in SEC analyses, the Hollingshead Index was used as a multidimensional composite encompassing educational attainment, occupational prestige, sex, and marital status (Hollingshead, 1975). This method assumes that the education and occupation of both spouses play an equal part in the SES of the family. Participants reported their age, sex, current household income, their own and their spouses’ (if married) education and occupations, or previous occupations if retired. A. B. Scott coded educational attainment and occupational prestige. Educational attainment was coded on a 7-point scale (i.e., 7 = graduate/professional training; 1 = less than 7th grade). Occupational prestige, defined as the level of assumed skill and power individuals possess in the maintenance functions of their position, was coded on a 9-point scale (e.g., 9 = higher executive, proprietor of large business, major professional; 1 = farm laborers, menial service workers). Following methods outlined by Hollingshead (1975), education and occupation scores were multiplied by factor weights (3 and 5, respectively). The SES score for a married participant was then calculated by summing the education and occupation scores for the participant and their spouse and dividing by two. The SES score for an unmarried participant was calculated by summing the individual’s education and occupation scores. Crystallized intelligence The North American Adult Reading Test (NAART; Blair & Spreen, 1989) was used to estimate crystallized intelligence. The NAART is a reliable and valid estimate of premorbid intelligence, comparable to the Wechsler Adult Intelligence Scale-Revised. The NAART is thought to be relatively resistant to neurological damage, though the change may occur under certain contexts. Decline may occur for those with mild to severe dementia, particularly when language is comprised. Factors such as cognitive reserve, social class, and education may also ameliorate decline (Strauss, Sherman, & Spreen, 2006). The NAART requires participants to read aloud a list of 61 irregularly spelled words; participants are scored for correct pronunciation. The NAART provides regression equations that estimate full-scale IQ (FSIQ), verbal IQ, and performance IQ based on the number of pronunciation errors (Blair & Spreen, 1989). The FSIQ estimate was utilized in the present study. Visuomotor speed and task switching The Trail Making Test (TMT) parts A and B were used to assess visuomotor speed and task switching speed, respectively (Reitan, 1955). Additionally, task switching efficiency was assessed using the derived TMT Part B minus A index (Drane, Yuspeh, Huthwaite, & Klingler, 2002). In Part A, participants are timed while drawing lines connecting numbers in ascending order as quickly as possible (i.e., 1-2-3, etc.). Similarly, in Part B, participants draw lines switching between numbers and letters in ascending and alphabetical order (i.e., 1-A-2-B-3-C, etc.). The TMT taps visuomotor speed in Part A and adds an executive component, task switching, in Part B, as flexibly switching between numbers and letters is required. However, as both parts A and B rely on visuomotor speed, the difference in completion time between the two parts (B minus A) may be used as a derived index to isolate the executive component of Part B, termed task switching efficiency. Normed scores based on age were used (Drane et al., 2002). Verbal memory The AVLT (Strauss et al., 2006) assessed several components of verbal memory including verbal learning, immediate recall, delayed recall, and recognition memory. The AVLT includes a 15-word list (List A), which was read to participants five times; participants were asked to recall as many words as possible after each trial (verbal learning; sum of A1–A5). Next, an interference list of 15 different words (List B) was read, and participants were asked to recall as many words as possible. After the recall of List B, participants were asked to recall words from List A (immediate recall; A6). Delayed recall of List A was assessed 20 min after immediate recall (delayed recall; A7), followed by a recognition trial including both target (recognition memory; List A) and distractor words (Strauss et al., 2006). Normed scores based on age were used (Ivnik et al., 1990). Data Analysis To describe SEC trajectories, multilevel models were used with years at Level 1 and people at Level 2 in SAS (9.4) PROC MIXED. Null models with restricted maximum likelihood estimation provided estimates of variance components and the intraclass correlation (ICC). Higher ICC values indicate more of the variance is due to stable between-person individual differences and less is due to within-person change over time. Growth models with maximum likelihood estimation and between-within degrees of freedom included the linear and quadratic effects of age as fixed effects. Fixed effects of age are reported as gamma weights, which are analogous to unstandardized beta weights in regression. The Level 1 equation is presented below: SECij=β0j+β1j(Agei) +β2j(Age2i) +εij SEC trajectories for person j at year i are a function of an intercept (SEC at the first time point), a linear and quadratic age slope, and a within-person residual. Random effects were included when indicated by the likelihood ratio test with mixture degrees of freedom. Specifically, random effects of intercept and linear age were included, but the quadratic effect of age was uniform across the sample and therefore not included. The Level 2 equation is presented below: β0j=γ00+ U0j β1j=γ10+ U1j β2j=γ20 Thus, with substitution, this model can be written as: SECij=γ00+ U0j+ Age (γ10+ U1j) + Age2(γ20) +εij To test our hypotheses, we employed two sets of hierarchical regressions in SPSS (24.0). Specifically, we regressed the neurocognitive variables on (1) SEC in childhood and adulthood, controlling for current SES, and (2) residential mobility in childhood and adulthood. A third set of exploratory hierarchical regressions evaluated associations between SEC in childhood and in developmental periods within adulthood, namely young adulthood (ages 19–39) and midlife (ages 40–60), on later-life cognition. Last, we conducted post hoc sensitivity analyses that further controlled for childhood SES in the SEC analyses. Two items from the Childhood Trauma Questionnaire (CTQ; Bernstein et al., 2003) were included to broadly account for childhood material wealth. The two items were “I didn’t have enough to eat” and “I had to wear dirty clothes”; participants responded on a 5-point scale from “never true” to “often true”. In general, this sample did not have severe deprivation in childhood; the majority of participants endorsed “never true” for both items (not enough to eat: 80%; wore dirty clothes: 82%), and the remaining endorsed “rarely true” (not enough to eat: 6%; wore dirty clothes: 5%), “sometimes true” (not enough to eat: 4%; wore dirty clothes: 4%); and “often true” (not enough to eat: 1%; wore dirty clothes: 0%). CTQ data for 11 participants (9%) were missing. Because some participants (N = 56) were administered the TMT and AVLT in an earlier study phase, a dummy variable was included to account for practice effects (1 = previously tested; 0 = not previously tested). Last, we did not include age in the models because crystallized intelligence remains largely intact into old age (e.g., Christensen, 2001) and because the neurocognitive scores for visuomotor speed and task switching (TMT) and verbal memory (AVLT) were already normed based on age. The Benjamini and Hochberg (1995) procedure was applied to the results to correct for multiple comparisons. Significance levels correcting for the false discovery rate (FDR) of 0.10 were calculated (a) for each hierarchical model/step and (b) for each predictor across hierarchical models separated into three families of outcomes, representing crystallized intelligence, visuomotor processing (three variables: visuomotor speed, task switching, and task switching efficiency), and verbal memory (four variables: total learning, immediate recall, delayed recall, and recognition). Results Lifetime Residential Addresses: Descriptive Statistics The majority of lifetime residential addresses (63% person-year observations) were in the Commonwealth of Kentucky. Within Kentucky, there were 46 different counties represented, with variability in the SEC values across these counties (M = 0.25, SD = 0.58, range = −2.81 to 0.99). The remaining addresses (37% person-year observations) were from 198 different counties outside of Kentucky, which had slightly more SEC variability (M = 0.32, SD = 0.67, range = −1.83 to 3.10). Table 1 depicts descriptive statistics and correlations among demographic data, SEC and SES parameters, and residential mobility. SEC was significantly lower [t(113)= −7.38, p < .001] in childhood (M = −0.048, SD = 0.66) than adulthood (M = 0.42, SD = 0.29). In exploratory models, we also examined SEC in young adulthood (19–39 years) and midlife (40–60 years). SEC was significantly lower [t(113)= −2.56, p = 0.012) in young adulthood (M = 0.38, SD = 0.37) than in midlife (M = 0.46, SD = 0.30). Most variation in lifetime (birth to age 60) SEC was due to within-person changes (ICC = .25). Adulthood SEC (ICC = .35) had more within-person variability than childhood SEC (ICC = .76), which was more stable. Additionally, young adulthood SEC had more within-person variability (ICC = .40) than midlife SEC (ICC = .60). Growth models indicated that SEC increased slightly with age (linear γ = .01, SE = .002, p< .001), with slower increases in older age (quadratic γ= −.0002, SE = .00002, p < .001). A statistically significant (p < .001) random effect of age indicated that there were individual differences in SEC trajectories. However, examination of Figure 1 suggests that most of the variation in SEC was due to nonlinear changes in the level of exposure over time rather than linear trends across the lifespan. On average, participants moved twice (SD = 1.97, range = 0–9 moves) during childhood (ages 0–18) and seven times (SD = 3.02, range = 0–16) during adulthood (ages 19–60 years). Table 1. Means of and Correlations Between Age, Gender, SEC, SES, and Residential Mobility Variable Mean (SD) Correlations 1 2 3 4 5 6 7 8 9 10 11 1. Age 75 (5.9) 2. Gender 60% female −.06 3. Childhood SEC −0.06 (0.66) −.01 .17 4. Adulthood SEC 0.42 (0.29) .15 −.06 .19* 5. Young adulthood SEC 0.38 (0.37) .14 .01 .35** .90** 6. Middle adulthood SEC 0.46 (0.30) .13 −.12 −.07 .84** .51** 7. Childhood SES (enough to eat) 0.19 (0.55) .08 −.07 −.11 .07 .04 .09 8. Childhood SES (wore dirty clothes) 0.13 (0.43) .17 −.15 −.02 −.11 −.13 −.06 .33** 9. Current SES 55.2 (7.9) .10 .01 −.04 −.13 −.11 −.11 −.08 .04 10. Moves in childhood 2 (2.0) −.05 .05 .04 .13 .16 .05 .13 .04 .07 11. Moves in adulthood 7 (3.0) .24** .03 −.02 −.10 −.13 −.03 .10 .22* .19* −.05 12. Total moves 9 (3.6) .17 .02 .02 −.02 −.03 .01 .15 .20* .18* .52** .83** Variable Mean (SD) Correlations 1 2 3 4 5 6 7 8 9 10 11 1. Age 75 (5.9) 2. Gender 60% female −.06 3. Childhood SEC −0.06 (0.66) −.01 .17 4. Adulthood SEC 0.42 (0.29) .15 −.06 .19* 5. Young adulthood SEC 0.38 (0.37) .14 .01 .35** .90** 6. Middle adulthood SEC 0.46 (0.30) .13 −.12 −.07 .84** .51** 7. Childhood SES (enough to eat) 0.19 (0.55) .08 −.07 −.11 .07 .04 .09 8. Childhood SES (wore dirty clothes) 0.13 (0.43) .17 −.15 −.02 −.11 −.13 −.06 .33** 9. Current SES 55.2 (7.9) .10 .01 −.04 −.13 −.11 −.11 −.08 .04 10. Moves in childhood 2 (2.0) −.05 .05 .04 .13 .16 .05 .13 .04 .07 11. Moves in adulthood 7 (3.0) .24** .03 −.02 −.10 −.13 −.03 .10 .22* .19* −.05 12. Total moves 9 (3.6) .17 .02 .02 −.02 −.03 .01 .15 .20* .18* .52** .83** Notes: Gender is coded 1 = male, 2 = female. Childhood SEC (ages 0–18); adulthood SEC (ages 19–60); young adulthood SEC (ages 19–39); middle adulthood SEC (ages 40–60). SEC = socioeconomic context; SES = socioeconomic status. *p < .05. **p < .01. View Large Table 1. Means of and Correlations Between Age, Gender, SEC, SES, and Residential Mobility Variable Mean (SD) Correlations 1 2 3 4 5 6 7 8 9 10 11 1. Age 75 (5.9) 2. Gender 60% female −.06 3. Childhood SEC −0.06 (0.66) −.01 .17 4. Adulthood SEC 0.42 (0.29) .15 −.06 .19* 5. Young adulthood SEC 0.38 (0.37) .14 .01 .35** .90** 6. Middle adulthood SEC 0.46 (0.30) .13 −.12 −.07 .84** .51** 7. Childhood SES (enough to eat) 0.19 (0.55) .08 −.07 −.11 .07 .04 .09 8. Childhood SES (wore dirty clothes) 0.13 (0.43) .17 −.15 −.02 −.11 −.13 −.06 .33** 9. Current SES 55.2 (7.9) .10 .01 −.04 −.13 −.11 −.11 −.08 .04 10. Moves in childhood 2 (2.0) −.05 .05 .04 .13 .16 .05 .13 .04 .07 11. Moves in adulthood 7 (3.0) .24** .03 −.02 −.10 −.13 −.03 .10 .22* .19* −.05 12. Total moves 9 (3.6) .17 .02 .02 −.02 −.03 .01 .15 .20* .18* .52** .83** Variable Mean (SD) Correlations 1 2 3 4 5 6 7 8 9 10 11 1. Age 75 (5.9) 2. Gender 60% female −.06 3. Childhood SEC −0.06 (0.66) −.01 .17 4. Adulthood SEC 0.42 (0.29) .15 −.06 .19* 5. Young adulthood SEC 0.38 (0.37) .14 .01 .35** .90** 6. Middle adulthood SEC 0.46 (0.30) .13 −.12 −.07 .84** .51** 7. Childhood SES (enough to eat) 0.19 (0.55) .08 −.07 −.11 .07 .04 .09 8. Childhood SES (wore dirty clothes) 0.13 (0.43) .17 −.15 −.02 −.11 −.13 −.06 .33** 9. Current SES 55.2 (7.9) .10 .01 −.04 −.13 −.11 −.11 −.08 .04 10. Moves in childhood 2 (2.0) −.05 .05 .04 .13 .16 .05 .13 .04 .07 11. Moves in adulthood 7 (3.0) .24** .03 −.02 −.10 −.13 −.03 .10 .22* .19* −.05 12. Total moves 9 (3.6) .17 .02 .02 −.02 −.03 .01 .15 .20* .18* .52** .83** Notes: Gender is coded 1 = male, 2 = female. Childhood SEC (ages 0–18); adulthood SEC (ages 19–60); young adulthood SEC (ages 19–39); middle adulthood SEC (ages 40–60). SEC = socioeconomic context; SES = socioeconomic status. *p < .05. **p < .01. View Large Childhood and adulthood SEC were modestly correlated (r = .19, p = .042); however, childhood SEC was related to young adulthood SEC (r = .35, p < .001) but not midlife SEC (r = −.07, p = .47). Also note that SES and SEC, whose distinct effects we are examining, were uncorrelated (current SES and adulthood SEC: r = −.13, p = .17; current SES and childhood SEC: r = −.04, p = .72; childhood SES and childhood SEC: rs = −.11 to −.02, ps = .25−.82). Childhood and adulthood residential mobility were uncorrelated (r = −.05, p = .61). Residential mobility in adulthood was positively correlated with age (r = .24, p = .009). SEC and residential mobility were not strongly related, but higher current SES was modestly correlated with more residential mobility in adulthood (r = .18, p = .049). Socioeconomic Context and Cognition Regarding the influences of broad socioeconomic environmental conditions, we hypothesized that higher SEC, particularly in the early developmental years, would be associated with better later-life cognition. Analyses included current SES in the final step of hierarchical regression models to examine whether SEC effects remained above and beyond current SES. Table 2 presents hierarchical regression results. In line with our hypothesis, higher childhood SEC was associated with better recognition memory (β = .240, p = .012). This effect remained when childhood SEC was the only main effect (Step 1), when adulthood SEC was added (Step 2), and when current SES was also added (Step 3). Higher adulthood SEC was significantly associated with better immediate recall (β = .259, p = .008) above and beyond current SES. Table 2. Hierarchical Regression Models of Child and Adult SEC on Late-Life Cognition Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 Β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027† .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a,b Step 2 .001 .029† .032† .020 .006 .065* .023 .005  Practice .164 .121 .087 .108 .052 .175 −.037  Child SEC .159 −.061 −.127 −.129 .079 .058 .000 .230*a,b  Adult SEC .030 .175† .182† .143 .080 .260*a,b .156 .073 Step 3 .048* .001 .000 .000 .001 .000 .001 .035*  Practice .170 .120 .084 .114 .051 .181 .003  Child SEC .161† −.060 −.128 −.130 .078 .058 −.001 .230*b  Adult SEC .058 .180† .181† .141 .085 .259*a,b .160 .102  Current SES .220*a,b .037 −.010 −.018 .031 −.005 .028 .192* Total R2 .076* .055 .056 .039 .024 .076 .051 .100* n 112 112 110 110 110 110 110 110 Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 Β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027† .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a,b Step 2 .001 .029† .032† .020 .006 .065* .023 .005  Practice .164 .121 .087 .108 .052 .175 −.037  Child SEC .159 −.061 −.127 −.129 .079 .058 .000 .230*a,b  Adult SEC .030 .175† .182† .143 .080 .260*a,b .156 .073 Step 3 .048* .001 .000 .000 .001 .000 .001 .035*  Practice .170 .120 .084 .114 .051 .181 .003  Child SEC .161† −.060 −.128 −.130 .078 .058 −.001 .230*b  Adult SEC .058 .180† .181† .141 .085 .259*a,b .160 .102  Current SES .220*a,b .037 −.010 −.018 .031 −.005 .028 .192* Total R2 .076* .055 .056 .039 .024 .076 .051 .100* n 112 112 110 110 110 110 110 110 Note: FDR = false discovery rate; SEC = socioeconomic context; SES = socioeconomic status. aStatistically significant after FDR correction applied within each hierarchical step. bStatistically significant after FDR correction applied within each predictor. †p < .10. *p < .05. View Large Table 2. Hierarchical Regression Models of Child and Adult SEC on Late-Life Cognition Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 Β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027† .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a,b Step 2 .001 .029† .032† .020 .006 .065* .023 .005  Practice .164 .121 .087 .108 .052 .175 −.037  Child SEC .159 −.061 −.127 −.129 .079 .058 .000 .230*a,b  Adult SEC .030 .175† .182† .143 .080 .260*a,b .156 .073 Step 3 .048* .001 .000 .000 .001 .000 .001 .035*  Practice .170 .120 .084 .114 .051 .181 .003  Child SEC .161† −.060 −.128 −.130 .078 .058 −.001 .230*b  Adult SEC .058 .180† .181† .141 .085 .259*a,b .160 .102  Current SES .220*a,b .037 −.010 −.018 .031 −.005 .028 .192* Total R2 .076* .055 .056 .039 .024 .076 .051 .100* n 112 112 110 110 110 110 110 110 Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 Β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027† .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a,b Step 2 .001 .029† .032† .020 .006 .065* .023 .005  Practice .164 .121 .087 .108 .052 .175 −.037  Child SEC .159 −.061 −.127 −.129 .079 .058 .000 .230*a,b  Adult SEC .030 .175† .182† .143 .080 .260*a,b .156 .073 Step 3 .048* .001 .000 .000 .001 .000 .001 .035*  Practice .170 .120 .084 .114 .051 .181 .003  Child SEC .161† −.060 −.128 −.130 .078 .058 −.001 .230*b  Adult SEC .058 .180† .181† .141 .085 .259*a,b .160 .102  Current SES .220*a,b .037 −.010 −.018 .031 −.005 .028 .192* Total R2 .076* .055 .056 .039 .024 .076 .051 .100* n 112 112 110 110 110 110 110 110 Note: FDR = false discovery rate; SEC = socioeconomic context; SES = socioeconomic status. aStatistically significant after FDR correction applied within each hierarchical step. bStatistically significant after FDR correction applied within each predictor. †p < .10. *p < .05. View Large Turning to the influences of current individual socioeconomic position, higher current SES was significantly associated with higher estimated crystallized intelligence (β = .220, p = .019) and tended to be associated with better recognition memory (β = .192, p = .045), however this effect did not pass FDR corrections. These SES effects were above and beyond childhood and adulthood SEC effects. There were no significant associations between childhood SEC, adulthood SEC, or current SES with visuomotor speed, task switching speed or efficiency, verbal learning, or delayed recall. Higher childhood SEC tended to be associated with higher estimated crystallized intelligence (β = .161, p = .088). Additionally, higher adulthood SEC tended to be associated with better visuomotor speed (β = .180, p = .065) and faster task switching (β = .181, p = .066), however, these associations were not statistically significant. Exploratory analyses (Table 3) evaluated associations between SEC in childhood and in developmental periods within adulthood, namely young adulthood (ages 19–39) and midlife (ages 40–60), on later-life cognition. The aforementioned significant effects (Table 2) remained with the exception that associations between childhood SEC and recognition memory did not pass FDR corrections in Steps 2–4. Additional effects also emerged: when including childhood SEC and young adulthood SEC in the model, higher midlife SEC was significantly associated with faster task switching (β = .262, p = .025) and better task switching efficiency (β = .268, p = .022). These results remained significant when current SES was included in the model (Step 4) and after FDR correction applied within each predictor. Additionally, higher young adulthood SEC was associated with better immediate recall (β = .265, p = .009). This effect decreased and became nonsignificant in Steps 3 (including midlife SEC, β = .214, p = .082) and 4 (including current SES, β = .213, p = .085). Young adulthood SEC was not significantly associated with any other later-life cognition variables. Table 3. Hierarchical Regression Models of Child, Young Adult (YA), and Middle Adult (MA) SEC on Later-Life Cognition Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027 .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a Step 2 .005 .022 .010 .003 .009 .062* .024 .003  Practice .157 .111 .079 .106 .043 .170 −.040  Child SEC .139 −.085 −.132 −.124 .059 .013 −.029 .221*  YA SEC .074 .160 .106 .058 .099 .265*a .164 .056 Step 3 .006 .008 .045* .047* .000 .005 .001 .003  Practice .165 .133 .101 .105 .049 .173 −.035  Child SEC .111 −.054 −.054 −.044 .054 .038 −.018 .242*  YA SEC .133 .095 −.054 −.106 .109 .214† .141 .012  MA SEC −.096 .107 .262*b .268*b −.016 .084 .039 .071 Step 4 .047* .001 .000 .000 .001 .003 .001 .003*  Practice .172 .132 .098 .111 .048 .179 .005  Child SEC .115 −.052 −.054 −.044 .054 .038 −.018 .240*  YA SEC .147 .097 −.054 −.107 .112 .213† .143 .031  MA SEC −.079 .111 .261*b .267*b −.014 .083 .041 .085  Current SES .219*a,b .037 −.007 −.014 .031 −.005 .028 .192* Total R2 .085* .056 .079 .070 .026 .077 .052 .103* n 112 112 110 110 110 110 110 110 Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027 .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a Step 2 .005 .022 .010 .003 .009 .062* .024 .003  Practice .157 .111 .079 .106 .043 .170 −.040  Child SEC .139 −.085 −.132 −.124 .059 .013 −.029 .221*  YA SEC .074 .160 .106 .058 .099 .265*a .164 .056 Step 3 .006 .008 .045* .047* .000 .005 .001 .003  Practice .165 .133 .101 .105 .049 .173 −.035  Child SEC .111 −.054 −.054 −.044 .054 .038 −.018 .242*  YA SEC .133 .095 −.054 −.106 .109 .214† .141 .012  MA SEC −.096 .107 .262*b .268*b −.016 .084 .039 .071 Step 4 .047* .001 .000 .000 .001 .003 .001 .003*  Practice .172 .132 .098 .111 .048 .179 .005  Child SEC .115 −.052 −.054 −.044 .054 .038 −.018 .240*  YA SEC .147 .097 −.054 −.107 .112 .213† .143 .031  MA SEC −.079 .111 .261*b .267*b −.014 .083 .041 .085  Current SES .219*a,b .037 −.007 −.014 .031 −.005 .028 .192* Total R2 .085* .056 .079 .070 .026 .077 .052 .103* n 112 112 110 110 110 110 110 110 Note: FDR = false discovery rate; MA = middle adult; SEC = socioeconomic context; SES = socioeconomic status; YA = young adulthood. aStatistically significant after FDR correction applied within each hierarchical step. bStatistically significant after FDR correction applied within each predictor. †p < .10. *p < .05. View Large Table 3. Hierarchical Regression Models of Child, Young Adult (YA), and Middle Adult (MA) SEC on Later-Life Cognition Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027 .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a Step 2 .005 .022 .010 .003 .009 .062* .024 .003  Practice .157 .111 .079 .106 .043 .170 −.040  Child SEC .139 −.085 −.132 −.124 .059 .013 −.029 .221*  YA SEC .074 .160 .106 .058 .099 .265*a .164 .056 Step 3 .006 .008 .045* .047* .000 .005 .001 .003  Practice .165 .133 .101 .105 .049 .173 −.035  Child SEC .111 −.054 −.054 −.044 .054 .038 −.018 .242*  YA SEC .133 .095 −.054 −.106 .109 .214† .141 .012  MA SEC −.096 .107 .262*b .268*b −.016 .084 .039 .071 Step 4 .047* .001 .000 .000 .001 .003 .001 .003*  Practice .172 .132 .098 .111 .048 .179 .005  Child SEC .115 −.052 −.054 −.044 .054 .038 −.018 .240*  YA SEC .147 .097 −.054 −.107 .112 .213† .143 .031  MA SEC −.079 .111 .261*b .267*b −.014 .083 .041 .085  Current SES .219*a,b .037 −.007 −.014 .031 −.005 .028 .192* Total R2 .085* .056 .079 .070 .026 .077 .052 .103* n 112 112 110 110 110 110 110 110 Cognitive construct Crystallized intelligence Visuomotor speed Task switch speed Task switch efficiency Verbal learning Immediate recall Delayed recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .027 .025 .024 .019 .017 .011 .027 .062*  Practice .150 .107 .077 .102 .034 .165 −.042  Child SEC .165† −.030 −.095 −.104 .093 .105 .028 .240*a Step 2 .005 .022 .010 .003 .009 .062* .024 .003  Practice .157 .111 .079 .106 .043 .170 −.040  Child SEC .139 −.085 −.132 −.124 .059 .013 −.029 .221*  YA SEC .074 .160 .106 .058 .099 .265*a .164 .056 Step 3 .006 .008 .045* .047* .000 .005 .001 .003  Practice .165 .133 .101 .105 .049 .173 −.035  Child SEC .111 −.054 −.054 −.044 .054 .038 −.018 .242*  YA SEC .133 .095 −.054 −.106 .109 .214† .141 .012  MA SEC −.096 .107 .262*b .268*b −.016 .084 .039 .071 Step 4 .047* .001 .000 .000 .001 .003 .001 .003*  Practice .172 .132 .098 .111 .048 .179 .005  Child SEC .115 −.052 −.054 −.044 .054 .038 −.018 .240*  YA SEC .147 .097 −.054 −.107 .112 .213† .143 .031  MA SEC −.079 .111 .261*b .267*b −.014 .083 .041 .085  Current SES .219*a,b .037 −.007 −.014 .031 −.005 .028 .192* Total R2 .085* .056 .079 .070 .026 .077 .052 .103* n 112 112 110 110 110 110 110 110 Note: FDR = false discovery rate; MA = middle adult; SEC = socioeconomic context; SES = socioeconomic status; YA = young adulthood. aStatistically significant after FDR correction applied within each hierarchical step. bStatistically significant after FDR correction applied within each predictor. †p < .10. *p < .05. View Large Post hoc sensitivity analyses further controlled for childhood SES proxies and evaluated associations between (a) SEC in childhood and adulthood on later-life cognition (Supplementary Table S1) and (b) SEC in childhood, young adulthood, and midlife on later-life cognition (Supplementary Table S2). The aforementioned significant effects remained, and no additional effects emerged. Residential Mobility and Cognition Last, we hypothesized that less residential mobility would be associated with better later-life cognition. Table 4 presents hierarchical regression results. Contrary to our hypothesis, significant effects were in the direction of higher residential mobility associating with better later-life cognition. Specifically, higher residential mobility in childhood (but not adulthood) was significantly associated with higher estimated crystallized intelligence (childhood: β = .194, p = .040; adulthood: β = .077, p = .42). Higher residential mobility in childhood tended to be associated with better immediate (β = .197, p = .040) and delayed (β = .185, p = .048) recall, and higher residential mobility in adulthood tended to be associated with better verbal learning (adulthood: β = .216, p = .029), however, these effects did not pass FDR corrections. Residential mobility in childhood and adulthood did not significantly relate to visuomotor speed, task switching speed or efficiency, or recognition memory. Table 4. Hierarchical Regression Models of Residential Mobility on Cognitive Functioning in Later Life Cognitive Construct Crystallized Intelligence Visuomotor Speed Task Switch Speed Task Switch Efficiency Verbal Learning Immediate Recall Delayed Recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .038* .023 .019 .013 .039 .040 .064* .015  Practice .150 .133 .108 .108 .023 .175 −.076  Moves in childhood .194*a,b −.030 −.033 −.043 .163† .197* .178† .099 Step 2 .006 .001 .019 .019 .042* .003 .029† .013  Practice .141 .089 .064 .171 .039 .228* −.041  Moves in childhood .198*a,b −.031 −.037 −.048 .171† .199* .185* .104  Moves in adulthood .077 −.029 −.145 −.146 .216* .054 .180† .121 Total R2 .043 .024 .038 .033 .082* .042 .093* .028 n 111 111 109 109 110 109 109 109 Cognitive Construct Crystallized Intelligence Visuomotor Speed Task Switch Speed Task Switch Efficiency Verbal Learning Immediate Recall Delayed Recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .038* .023 .019 .013 .039 .040 .064* .015  Practice .150 .133 .108 .108 .023 .175 −.076  Moves in childhood .194*a,b −.030 −.033 −.043 .163† .197* .178† .099 Step 2 .006 .001 .019 .019 .042* .003 .029† .013  Practice .141 .089 .064 .171 .039 .228* −.041  Moves in childhood .198*a,b −.031 −.037 −.048 .171† .199* .185* .104  Moves in adulthood .077 −.029 −.145 −.146 .216* .054 .180† .121 Total R2 .043 .024 .038 .033 .082* .042 .093* .028 n 111 111 109 109 110 109 109 109 Note: FDR = false discovery rate. aStatistically significant after FDR correction applied within each hierarchical step. bStatistically significant after FDR correction applied within each predictor. †p < .10. *p < .05. View Large Table 4. Hierarchical Regression Models of Residential Mobility on Cognitive Functioning in Later Life Cognitive Construct Crystallized Intelligence Visuomotor Speed Task Switch Speed Task Switch Efficiency Verbal Learning Immediate Recall Delayed Recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .038* .023 .019 .013 .039 .040 .064* .015  Practice .150 .133 .108 .108 .023 .175 −.076  Moves in childhood .194*a,b −.030 −.033 −.043 .163† .197* .178† .099 Step 2 .006 .001 .019 .019 .042* .003 .029† .013  Practice .141 .089 .064 .171 .039 .228* −.041  Moves in childhood .198*a,b −.031 −.037 −.048 .171† .199* .185* .104  Moves in adulthood .077 −.029 −.145 −.146 .216* .054 .180† .121 Total R2 .043 .024 .038 .033 .082* .042 .093* .028 n 111 111 109 109 110 109 109 109 Cognitive Construct Crystallized Intelligence Visuomotor Speed Task Switch Speed Task Switch Efficiency Verbal Learning Immediate Recall Delayed Recall Recognition ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β ΔR2 β Step 1 .038* .023 .019 .013 .039 .040 .064* .015  Practice .150 .133 .108 .108 .023 .175 −.076  Moves in childhood .194*a,b −.030 −.033 −.043 .163† .197* .178† .099 Step 2 .006 .001 .019 .019 .042* .003 .029† .013  Practice .141 .089 .064 .171 .039 .228* −.041  Moves in childhood .198*a,b −.031 −.037 −.048 .171† .199* .185* .104  Moves in adulthood .077 −.029 −.145 −.146 .216* .054 .180† .121 Total R2 .043 .024 .038 .033 .082* .042 .093* .028 n 111 111 109 109 110 109 109 109 Note: FDR = false discovery rate. aStatistically significant after FDR correction applied within each hierarchical step. bStatistically significant after FDR correction applied within each predictor. †p < .10. *p < .05. View Large Discussion The present study utilized a novel method of measuring the relative wealth and quality of the environment, or SEC, to predict cognitive functioning among generally healthy older adults. Census data on county-level unemployment rate, median household income, and percent of individuals with a high school diploma were obtained from reported lifetime residential addresses to compute SEC. Our primary focus was on the influences of broad socioeconomic environmental conditions, particularly in early life, on later-life cognition. Higher childhood SEC (ages 0–18), controlling for practice effects, adulthood SEC, and current SES, was significantly associated with better recognition memory and tended to be associated with higher estimated crystallized intelligence. Notably, of the verbal memory functions assessed in the present study, recognition memory is most reflective of retention and least impacted by executive influence on memory retrieval (i.e., not reliant on free recall; Lezak, 2004). Interestingly, in contrast with childhood SEC, adulthood SEC effects tended to be executive. Higher adulthood SEC (ages 19–60), controlling for practice effects, childhood SEC, and current SES, was significantly associated with better immediate verbal memory recall, whereas higher midlife SEC (ages 40–60) was significantly associated with faster task switching and better task switching efficiency. Immediate verbal memory requires free recall, which has an executive retrieval component (Lezak, 2004). Therefore, in the broad community context, early socioeconomic contexts (childhood SEC) may impact the development of cognitive reserve and long-term integrity of verbal memory functioning, whereas more proximal socioeconomic environmental conditions (midlife SEC) may impact higher-level cognitive functions, which may be among the first to deteriorate with aging (Salthouse, Atkinson, & Berish, 2003). Together, these findings suggest there may be a quadratic function in the sensitive periods of SEC effects on later-life cognition. Importantly, these SEC findings were independent of current individual-level SES. Although the present investigation focused on broader environmental socioeconomic conditions, individual-level SES also affects later-life cognition. In line with previous evidence (Gottfredson, 2004), higher current SES (controlling for practice effects and SEC) was associated with higher estimated crystallized intelligence. This association likely reflects bidirectional relationships between crystallized intelligence with elements of SES including education and occupational status. Contrary to hypotheses, higher residential mobility in childhood was associated with better cognitive function in later-life, specifically higher estimated crystallized intelligence. For some cohorts, life stages, and subcultural contexts, an individual’s or family’s ability to move residences may reflect intellectual and socioeconomic resources that result in the freedom to change employment or residence as opposed to economic instability. In the present study, higher residential mobility in adulthood was associated with higher current SES. Our study contributes a new methodological approach to measure SEC. Advantages of this methodology are that objective address data are less likely to be misremembered (Ward, 2011) and unlikely to be influenced by a positivity bias in older adults (Carstensen, 2006). Conversely, subjective reports of socioeconomic position and adversity in childhood can be misremembered in a positive light, diminishing the accuracy of these estimates (Maughan & Rutter, 1997). Another advantage of this SEC approach is that multiple parameters can be derived from these data. In the present study, total SEC exposure in childhood and adulthood, as well as residential mobility, were calculated. The high stability of SEC before age 19 in the present sample made smaller distinctions (e.g., early vs middle childhood vs adolescence) unfeasible; however, we explored additional developmental periods within adulthood (i.e., young vs middle adulthood). It should be noted that SEC is particularly refined in the present sample with regard to individual exposures because the majority of lifetime residential addresses were in Kentucky. Kentucky has the fourth highest number of counties in the United States, with 120 counties each covering only an average of 337 square miles. Although there was sufficient SEC variability, it is important to examine this method in other geographic locations in the future. Additionally, the present study employed county-level census data to capture the broad environmental context. Future investigations might explore using other census measures such as census tracts (which were not uniformly collected across the United States until the 2000 Census and therefore not used in the present study) to capture the nuances of SEC exposure. An important consideration in the present study is whether SEC contributes new information to our understanding of social stratification and health above and beyond individual-level measures of SES. Although additional research is needed to empirically test this question, our results suggest that SEC and SES are nonoverlapping constructs that may provide unique information (e.g., correlations between SES and SEC were small and not significant; Table 1). Additionally, SEC effects on later life cognition were independent of current and childhood individual-level SES. SEC may reflect health-relevant exosystem-level factors, such as physical environmental exposures and access to health care, that affect later life cognition (e.g., Cherrie et al., 2018). To test this interpretation of the representativeness of our SEC variable, we conducted a post hoc analysis using county health rankings from the University of Wisconsin (UW) Population Health Institute (Remington, Catlin, & Gennuso, 2015). The UW data contain annual (2010–2017) county-level measures of “health by place” within two main domains: health outcomes (HO), which include quality of life, morbidity, and mortality data; and health factors (HF), which include socioeconomic factors (education, employment, income, social support, community safety/crime), access to and quality of healthcare, health behaviors (e.g., tobacco and drug use), and physical environmental factors (air and water quality, housing, and transportation). Lower county-level HO and HF values represent higher within-state rankings. We identified lifelong Kentucky residents in our data whose lifetime SEC scores (averaged across ages 0–60) were within the top and bottom 10% of the SEC distribution (n = 9 individuals representing 20 counties). For each of the nine participants, we averaged the HO and HF rankings across all counties s/he lived in. In post hoc analyses, higher SEC was significantly associated with better HO rankings (r = −.82, p = .007) and better HF rankings (r = −.94, p < .001). Taken together, our results suggest that SES and SEC represent distinct constructs and that our SEC measure may capture a range of exosystem-level variables that operate above individual-level SES. Although there was a range in the indicators of current SES, the present sample was majority Caucasian, relatively wealthy, and well educated. The sample also demonstrated above average estimated premorbid intelligence (NAART FSIQ M = 112.45, range = 87.24–127.02). These characteristics may limit the generalizability of the current findings. Lifespan SEC trajectories were equifinal in that all participants were living in a relatively affluent county at assessment (Figure 1). Insofar as high SES and concurrent high SEC protect against adverse effects of early experience on cognition (Wight et al., 2006), results from this sample may reflect only the influence of childhood SEC that was resistant to these buffers. Additionally, although the effects of childhood SEC on later life cognition were independent of childhood SES (assessed using two items from the CTQ that broadly indicate material deprivation and may inexactly measure childhood SES), future research should corroborate this using other indicators of childhood SES (e.g., parental education, employment, and/or income). Furthermore, future work is needed that simultaneously examines whether and how microsystem-level variables (e.g., individual-level SES and childhood enrichment factors, such as number of children’s books in the home or frequency of someone in the home reading aloud to the child, etc.) and exosystem-level variables (i.e., SEC, green spaces, etc.) individually and interactively influence later-life cognition. Regarding the feasibility of the SEC approach, there was a low percentage of missing SEC data in the present sample. However, it should be noted that participants may misremember the exact years they lived at a certain address. Finally, the present investigation represents a preliminary examination of SEC on later life cognition; although all reported analyses correspond to theoretically and empirically informed a priori hypotheses, multiple analyses were conducted. Therefore, these effects require replication in large, heterogeneous samples. SEC may be a useful adjunct to conventional measures of early life SES, which may be prone to biases when self-reported. The SEC measure may benefit from further refinement in terms of indicators and categorization of SEC periods (childhood vs adulthood). The present study focused on SEC total exposure and residential mobility, but other parameters could be derived (e.g., SEC range or individual standard deviation as reflections of variability in SEC environments). Additionally, separate components of SEC could be investigated if theorized to affect later life cognition in unique ways. For example, particular aspects of SEC (e.g., school zone ratings, proximity to libraries) during certain developmental periods (e.g., early childhood) may be more strongly associated with cognitive reserve development. Further research can more fully establish the validity of this approach and the relationship between SEC and later-life health and cognitive outcomes. In addition, how SEC gets “under the skull” to affect cognitive development is an important question. Some pathways may be relatively direct and only involve the individual and the exosystem (e.g., exposure to environmental toxins and pollutants), but others may involve interactions among ecological levels, such as interactions between exosystem employment levels and microsystem family environments (Bronfenbrenner, 1977). Mediational models may also be fruitful; for example, employment levels might impact individual cognitive development through school quality. Socioeconomic-related health disparities exist for cognitive decline, dementia, and Alzheimer’s disease (D. A. Evans et al., 1997; Yaffe et al., 2013; Zhang et al., 2016). Understanding the risk factors for these diseases in the context of broader environmental influences is a critical first step to developing effective ways to promote healthy cognitive aging. Measuring SEC opens the door for an ecological understanding of lifespan cognitive development, with implications for later-life cognitive function. Supplementary Material Supplementary data are available at The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences online. Funding This work was supported by the Dana Foundation and the National Institutes of Health (R01-AG026307, K02-033629, K99-AG056635, P30-AG028383, and CTSA-UL1TR000117). Conflict of Interest The authors report no conflicts of interest. Acknowledgments Author Contributions: A. B. Scott and R. G. Reed contributed equally to this work. S. C. Segerstrom designed the parent study, including the measures. A. B. Scott and S. C. Segerstrom planned the study, and with R. G. Reed, performed the data analyses and wrote the article. N. E. Garcia-Willingham provided consultation for the neuropsychological measures and contributed to writing the article. K. A. Lawrence contributed to data analyses and writing the article. References American Fact Finder . ( 2017 ). Annual estimates of the resident population for selected age groups by sex . Retrieved November 5, 2017 from https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?src=CF Barnes , L. L. , Mendes de Leon , C. F. , Wilson , R. S. , Bienias , J. L. , & Evans , D. A . ( 2004 ). Social resources and cognitive decline in a population of older African Americans and whites . Neurology , 63 , 2322 – 2326 . doi: https://doi.org/10.1212/01.wnl.0000147473.04043.b3 Benjamini , Y. , & Hochberg , Y . ( 1995 ). Controlling the false discovery rate: A practical and powerful approach to multiple testing . Journal of the Royal Statistical Society Series B (Methodological) , 57 , 289 – 300 . doi:10.2307/2346101 Berkman , L. F. , Seeman , T. E. , Albert , M. , Blazer , D. , Kahn , R. , Mohs , R. ,… Rowe , J . ( 1993 ). High, usual and impaired functioning in community-dwelling older men and women: Findings from the MacArthur Foundation Research Network on Successful Aging . Journal of Clinical Epidemiology , 46 , 1129 – 1140 . doi: https://doi.org/10.1016/0895-4356(93)90112-E Bernstein , D. P. , Stein , J. A. , Newcomb , M. D. , Walker , E. , Pogge , D. , Ahluvalia , T. ,… Zule , W . ( 2003 ). Development and validation of a brief screening version of the Childhood Trauma Questionnaire . Child Abuse & Neglect , 27 , 169 – 190 . doi: https://doi.org/10.1016/S0145-2134(02)00541-0 Blair , J. R. , & Spreen , O . ( 1989 ). Predicting premorbid IQ: A revision of the national adult reading test . Clinical Neuropsychologist , 3 , 129 – 136 . doi: https://doi.org/10.1080/13854048908403285 Bronfenbrenner , U . ( 1977 ). Toward an experimental ecology of human development . American Psychologist , 32 , 513 – 531 . doi: https://doi.org/10.1037//0003-066x.32.7.513 Carstensen , L. L . ( 2006 ). The influence of a sense of time on human development . Science , 312 , 1913 – 1915 . doi: https://doi.org/10.1126/science.1127488 Cassarino , M. , O’Sullivan , V. , Kenny , R. A. , & Setti , A . ( 2016 ). Environment and cognitive aging: A cross-sectional study of place of residence and cognitive performance in the Irish longitudinal study on aging . Neuropsychology , 30 , 543 – 557 . doi: https://doi.org/10.1037/neu0000253 Charles , S. T. , Mather , M. , & Carstensen , L. L . ( 2003 ). Aging and emotional memory: The forgettable nature of negative images for older adults . Journal of Experimental Psychology , 132 , 310 – 324 . doi: https://doi.org/10.1037/0096-3445.132.2.310 Cherrie , M. P. , Shortt , N. K. , Mitchell , R. J. , Taylor , A. M. , Redmond , P. , Thompson , C. W. ,… Pearce , J. R . ( 2018 ). Green space and cognitive ageing: A retrospective life course analysis in the Lothian Birth Cohort 1936 . Social Science & Medicine (1982) , 196 , 56 – 65 . doi: https://doi.org/10.1016/j.socscimed.2017.10.038 Christensen , H . ( 2001 ). What cognitive changes can be expected with normal ageing ? The Australian and New Zealand Journal of Psychiatry , 35 , 768 – 775 . doi: https://doi.org/10.1046/j.1440-1614.2001.00966.x Drane , D. L. , Yuspeh , R. L. , Huthwaite , J. S. , & Klingler , L. K . ( 2002 ). Demographic characteristics and normative observations for derived-trail making test indices . Neuropsychiatry, Neuropsychology, and Behavioral Neurology , 15 , 39 – 43 . Duncan , G. J. , Brooks-Gunn , J. , & Klebanov , P. K . ( 1994 ). Emotion deprivation and early childhood development . Child Development , 65 , 296 – 318 . doi: https://doi.org/10.2307/1131385 Evans , D. A. , Hebert , L. E. , Beckett , L. A. , Scherr , P. A. , Albert , M. S. ,… & Taylor , J. O . ( 1997 ). Education and other measures of socioeconomic status and risk of incident Alzheimer disease in a defined population of older persons . Archives of Neurology , 54 , 1399 – 1405 . doi: https://doi.org/10.1001/archneur.1997.00550230066019 Evans , G. W. , & Kantrowitz , E . ( 2002 ). Socioeconomic status and health: The potential role of environmental risk exposure . Annual Review of Public Health , 23 , 303 – 331 . doi: https://doi.org/10.1146/annurev.publhealth.23.112001.112349 Fotenos , A. F. , Mintun , M. A. , Snyder , A. Z. , Morris , J. C. , & Buckner , R. L . ( 2008 ). Brain volume decline in aging: Evidence for a relation between socioeconomic status, preclinical Alzheimer disease, and reserve . Archives of Neurology , 65 , 113 – 120 . doi: https://doi.org/10.1001/archneurol.2007.27 Gottfredson , L. S . ( 2004 ). Intelligence: Is it the epidemiologists’ elusive “fundamental cause” of social class inequalities in health ? Journal of Personality and Social Psychology , 86 , 174 – 199 . doi: https://doi.org/10.1037/0022-3514.86.1.174 Hackman , D. A. , Farah , M. J. , & Meaney , M. J . ( 2010 ). Socioeconomic status and the brain: Mechanistic insights from human and animal research . Nature Reviews. Neuroscience , 11 , 651 – 659 . doi: https://doi.org/10.1038/nrn2897 Hollingshead , A. B . ( 1975 ). Four factor index of social status . New Haven, CT : Yale University . Unpublished manuscript. Ivnik , R. J. , Malec , J. F. , Tangalos , E. G. , Petersen , R. C. , Kokmen , E. , & Kurland , L. T . ( 1990 ). The Auditory-Verbal Learning Test (AVLT): Norms for ages 55 years and older . Psychological Assessment , 2 , 304 – 312 . doi: https://doi.org/10.1037//1040-3590.2.3.304 Jelleyman , T. , & Spencer , N . ( 2007 ). Residential mobility in childhood and health outcomes: A systematic review . Journal of Epidemiology and Community Health , 62 , 584 – 592 . doi: https://doi.org/10.1136/jech.2007.060103 Jiang , T. , Yu , J. T. , Tian , Y. , & Tan , L . ( 2013 ). Epidemiology and etiology of Alzheimer’s disease: From genetic to non-genetic factors . Current Alzheimer Research , 10 , 852 – 867 . doi: https://doi.org/10.2174/15672050113109990155 Kaplan , G. A. , Turrell , G. , Lynch , J. W. , Everson , S. A. , Helkala , E. L. , & Salonen , J. T . ( 2001 ). Childhood socioeconomic position and cognitive function in adulthood . International Journal of Epidemiology , 30 , 256 – 263 . doi: https://doi.org/10.1093/ije/30.2.256 Kauhanen , L. , Lakka , H. M. , Lynch , J. W. , & Kauhanen , J. ( 2006 ). Social disadvantages in childhood and risk of all-cause death and cardiovascular disease in later life: A comparison of historical and retrospective childhood information. International Journal of Epidemiology, 35, 962–968. doi: https://doi.org/10.1093/ije/dyl046 Krieger , N . ( 2001 ). A glossary for social epidemiology . Journal of Epidemiology & Community Health , 55 , 693 – 700 . doi: https://doi.org/10.1136/jech.55.10.693 Krieger , N. , Rehkopf , D. H. , Chen , J. T. , Waterman , P. D. , Marcelli , E. , & Kennedy , M . ( 2008 ). The fall and rise of US inequities in premature mortality: 1960–2002 . PLoS Medicine , 5 , e46 . doi: https://doi.org/10.1371/journal.pmed.0050046 Krieger , N. , Williams , D. R. , & Moss , N. E . ( 1997 ). Measuring social class in US public health research: Concepts, methodologies, and guidelines . Annual Review of Public Health , 18 , 341 – 378 . doi: https://doi.org/10.1146/annurev.publhealth.18.1.341 Lanphear , B. P . ( 2015 ). The impact of toxins on the developing brain . Annual Review of Public Health , 36 , 211 – 230 . doi: https://doi.org/10.1146/annurev-publhealth-031912-114413 Lezak , M. D . ( 2004 ). Neuropsychological assessment . New York, NY: Oxford University Press . Luo , Y. , & Waite , L. J . ( 2005 ). The impact of childhood and adult SES on physical, mental, and cognitive well-being in later life . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 60 , 93 – 101 . doi: https://doi.org/10.1093/geronb/60.2.s93 Lupien , S. J. , McEwen , B. S. , Gunnar , M. R. , & Heim , C . ( 2009 ). Effects of stress throughout the lifespan on the brain, behaviour and cognition . Nature Reviews. Neuroscience , 10 , 434 – 445 . doi: https://doi.org/10.1038/nrn2639 Maughan , B. , & Rutter , M . ( 1997 ). Retrospective reporting of childhood adversity: Issues in assessing long-term recall . Journal of Personality Disorders , 11 , 19 – 33 . doi:10.1521/pedi.1997.11.1.19 McLeod , J. D. , & Kessler , R. C . ( 1990 ). Socioeconomic status differences in vulnerability to undesirable life events . Journal of Health and Social Behavior , 31 , 162 – 172 . doi: https://doi.org/10.2307/2137170 Miller , G. E. , Chen , E. , & Parker , K. J . ( 2011 ). Psychological stress in childhood and susceptibility to the chronic diseases of aging: Moving toward a model of behavioral and biological mechanisms . Psychological Bulletin , 137 , 959 – 997 . doi: https://doi.org/10.1037/a0024768 Reitan , R. M . ( 1955 ). The relationship of the Trail Making Test to organic brain damage . Journal of Consulting Psychology , 19 , 393 – 394 . doi: https://doi.org/10.1037/h0044509 Remington , P. L. , Catlin , B. B. , & Gennuso , K. P . ( 2015 ). The county health rankings: Rationale and methods . Population Health Metrics , 13 , 11 . doi: https://doi.org/10.1186/s12963-015-0044-2 Salthouse , T. A. , Atkinson , T. M. , & Berish , D. E . ( 2003 ). Executive functioning as a potential mediator of age-related cognitive decline in normal adults . Journal of Experimental Psychology. General , 132 , 566 – 594 . doi: https://doi.org/10.1037/0096-3445.132.4.566 Santos , D. N. , Assis , A. M. , Bastos , A. C. , Santos , L. M. , Santos , C. A. , Strina , A. ,… Barreto , M. L . ( 2008 ). Determinants of cognitive function in childhood: A cohort study in a middle income context . BMC Public Health , 8 , 202 . doi: https://doi.org/10.1186/1471-2458-8-202 Sattler , C. , Toro , P. , Schönknecht , P. , & Schröder , J . ( 2012 ). Cognitive activity, education and socioeconomic status as preventive factors for mild cognitive impairment and Alzheimer’s disease . Psychiatry Research , 196 , 90 – 95 . doi: https://doi.org/10.1016/j.psychres.2011.11.012 Straughen , J. K. , Caldwell , C. H. , Osypuk , T. L. , Helmkamp , L. , & Misra , D. P . ( 2013 ). Direct and proxy recall of childhood socio-economic position and health . Paediatric and Perinatal Epidemiology , 27 , 294 – 302 . doi: https://doi.org/10.1111/ppe.12045 Strauss , E. , Sherman , E. M. , & Spreen , O . ( 2006 ). A compendium of neuropsychological tests: Administration, norms, and commentary . New York : Oxford University Press . Thomas , T. L. , DiClemente , R. , & Snell , S . ( 2014 ). Overcoming the triad of rural health disparities: How local culture, lack of economic opportunity, and geographic location instigate health disparities . Health Education Journal , 73 , 285 – 294 . doi: https://doi.org/10.1177/0017896912471049 Ward , M. M . ( 2011 ). Concordance of sibling’s recall of measures of childhood socioeconomic position . BMC Medical Research Methodology , 11 , 147 . doi: https://doi.org/10.1186/1471-2288-11-147 Wight , R. G. , Aneshensel , C. S. , Miller-Martinez , D. , Botticello , A. L. , Cummings , J. R. , Karlamangla , A. S. , & Seeman , T. E . ( 2006 ). Urban neighborhood context, educational attainment, and cognitive function among older adults . American Journal of Epidemiology , 163 , 1071 – 1078 . doi: https://doi.org/10.1093/aje/kwj176 Wright , C. B. , Sacco , R. L. , Rundek , T. , Delman , J. , Rabbani , L. , & Elkind , M . ( 2006 ). Interleukin-6 is associated with cognitive function: The Northern Manhattan Study . Journal of Stroke and Cerebrovascular Diseases , 15 , 34 – 38 . doi: https://doi.org/10.1016/j.jstrokecerebrovasdis.2005.08.009 Yaffe , K. , Falvey , C. , Harris , T. B. , Newman , A. , Satterfield , S. , Koster , A. ,… Simonsick , E .; Health ABC Study . ( 2013 ). Effect of socioeconomic disparities on incidence of dementia among biracial older adults: Prospective study . BMJ (Clinical research ed.) , 347 , f7051 . doi:10.1136/bmj.f7051 Yang , Y. C. , Gerken , K. , Schorpp , K. , Boen , C. , & Harris , K. M . ( 2017 ). Early-life socioeconomic status and adult physiological functioning: A life course examination of biosocial mechanisms . Biodemography and Social Biology , 63 , 87 – 103 . doi: https://doi.org/10.1080/19485565.2017.1279536 Zhang , Z. , Hayward , M. D. , & Yu , Y. L . ( 2016 ). Life course pathways to racial disparities in cognitive impairment among older Americans . Journal of Health and Social Behavior , 57 , 184 – 199 . doi: https://doi.org/10.1177/0022146516645925 © The Author(s) 2018. Published by Oxford University Press on behalf of The Gerontological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Journal

The Journals of Gerontology Series B: Psychological Sciences and Social SciencesOxford University Press

Published: Jun 6, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off