Skip to main content
NCBI home page
The Journals of Gerontology Series B: Psychological Sciences and Social Sciences logoLink to The Journals of Gerontology Series B: Psychological Sciences and Social Sciences
. 2021 Jan 6;76(8):1489–1498. doi: 10.1093/geronb/gbab003

The Effects of Selective Survival on Black Adults’ Cognitive Development

Shyuan Ching Tan 1,✉, Alyssa A Gamaldo 1,2, Timothy Brick 1, Roland J Thorpe Jr 3, Jason C Allaire 4, Keith E Whitfield 5
Editor: Angela Gutchess
PMCID: PMC8436692  PMID: 33406264

Abstract

Objectives

The theory of selective survival suggests that possibly around 70–75 years of age, Blacks may display substantive changes in their pattern of cognitive decline. This study examined the age-graded pattern of cognitive decline within older Blacks by describing a trend that characterizes differences in the change of cognitive decline from ages 51.5 to 95.5, and hypothesized that this age-graded pattern is nonlinear.

Method

Utilizing 2 waves of longitudinal data from the Baltimore Study of Black Aging, this study used multilevel modeling to test whether the interaction between age and the 3-year study period (time between waves) had a positive effect on changes in inductive reasoning, declarative memory, working memory, and perceptual speed.

Results

A significant positive interaction between age and wave was found for inductive reasoning, demonstrating an age-grade pattern of change/decline in cognitive pattern for Blacks aged 51.5–95.4. Simple slope probing via the Johnson–Neyman Technique suggested that Black adults ~64 years and younger experienced significant decline in inductive reasoning across study time, whereas for those older than 63.71, the decline was nonsignificant. No significant age–wave interactions were found for declarative memory, working memory, or perceptual speed.

Discussion

Findings suggest a selective survival effect for inductive reasoning ability among Blacks. With decline evident so early, common cognitive intervention programs targeting adults 65+ may come too late for Blacks, signifying the importance and urgency for early health interventions and public policy designed to promote cognitive reserve.

Keywords: Age-graded pattern, Cognitive change, Early cognitive and educational intervention

While most studies found that racial/ethnic older adults had a higher risk for cognitive impairment compared to non-Hispanic Whites (Manly & Mayeux, 2004), some cross-racial comparisons, particularly between Blacks and Whites, have shown mixed results regarding differences in decline in global cognition (measured by the Telephone Interview for Cognitive Status and Mini-Mental State Examination) and in executive function (Weuve et al., 2018). In general, studies have shown that older Blacks had lower initial scores compared to older Whites, but they did not agree on the Blacks versus Whites rates (i.e., slopes) of decline (Alley et al., 2007; Weuve et al., 2018). For example, Lee and colleagues (2012) concluded that a 3-year average decline in general cognitive status for older (mean age 79 years) African Americans was faster than that for non-African Americans, while Weuve and colleagues (2018) found that a 5-year rate of decline in global cognition did not differ significantly between Blacks and Whites adults 65 years and older. Another study (Alley et al., 2007) looking at older Americans aged 70–103 revealed that racial differences in the rate of decline varied by cognitive tasks. Specifically, while there was no difference found for the decline rate for the Serial 7’s task (a task of working memory), Black participants had slower decline rate for word recall tasks.

This conflicting evidence suggests that our understanding of older individuals’ cognitive development is incomplete and not generalizable to all racial groups. One possible reason is that most cognitive aging studies modeling the slope of change/decline across adulthood were based on samples comprised predominantly of Whites (e.g., Alley et al.’s (2007) and Wilson et al.’s (2002) studies). With few cognitive aging studies including large samples of Black adults of various demographic characteristics (e.g., age, gender, and education), we have a limited understanding of older Black adults’ cognitive health and development. To better account for the interindividual differences in sociodemographic, health, and behaviors commonly observed within the Black communities (Whitfield et al., 2008), studies using a within-racial group approach have been conducted to further elucidate racially/ethnically relevant factors that might account for between-person differences in cognitive change solely among older Blacks. For example, Aiken-Morgan and colleagues (2018) examined the patterns of global cognitive stability and change in urban, community-dwelling Blacks aged 48–95. The authors discovered that over a 3-year period, older Blacks who had declining (from normal to mild cognitively impaired) cognitive status had fewer years of education; whereas those who maintained normal cognitive status had healthier lung function. Furthermore, any inference and conclusion regarding the cognitive development of racial minority groups based on existing conflicting literature are likely confounded by differences in participants’ age range between studies. Some studies consisted of participants from mid-50s to mid-80s (e.g., Goh et al., 2012), while others included participants from ages 70 to 100+ (e.g., Alley et al., 2007). If individuals’ rate of change/decline in performance score is contingent upon the age at which they entered a study for testing, having participants with varying ages at baseline makes findings generated from studies not well calibrated for consistency. Comparison based on these studies is unlikely to yield a definitive age-graded pattern of cognitive change/decline within Blacks.

Given the prior literature, it is warranted to incorporate scientific approaches that go beyond those utilized in prior studies to better understand cognitive aging in Blacks. By not pivoting towards this new direction, the scientific community is likely to continue to have an obscured understanding which can result in bias conclusions regarding cognitive functioning and changes within Blacks. For example, research implicating lower performance or faster decline among older Blacks had often led to diagnoses or classifications of older Blacks as cognitively impaired (see Aiken-Morgan et al., 2008 and Aiken-Morgan et al., 2010 for their review and summary of this literature). Borrowing from the theories of persistent inequalities (Bénabou, 1994), bias classification like this creates uncalled for social stratification, and stratification helps perpetuate the persistent inequalities and discriminations in our society and in many aspects of Black individuals’ life. Aiken-Morgan and colleagues (2008, 2010), in their effort to disentangle race and cognitive ability, examined the role of early education attainment in explaining racial differences. However, much remains to be learned about how older Blacks, as a group, experience cognitive change/decline at different times during their adulthood (specifically, at different ages when they enter a study for testing). Additionally, it is likely that intervention programs based upon narrowly supported conclusions about older Blacks’ cognitive changes at best will result in ineffective preventive efforts, but at worst, can further perpetuate inequalities and discriminations through strengthening a wrong perception about older Blacks’ cognitive ability. Lastly, limited effort to elucidate the scientific understanding of cognitive aging in Blacks will further perpetuate unnecessary and/or inadequate health procedures often experienced by Blacks (Evans et al., 2020; Grady & Edgar, 2003). These unwarranted health practices facilitate inequitable and uneconomical health care services that negatively impact Black adults and families as well as our societal health care system (LaVeist et al., 2011).

Hence, to continue the effort of scholars endorsing the within-racial group approach, this study utilized the Baltimore Study of Black Aging (BSBA), which consists of only Black adults from 50 to 95 years old at baseline. Primarily, to capitalize on the study’s two-wave (approximately 3 years apart) longitudinal data and its wide age range, we attempted to characterize an age-graded pattern of cognitive change/decline. That is, if the slope of change/decline in performance scores from Wave 1 to 2 was correlated with the age they entered and stayed in our study for testing, we should see an age-graded pattern if all participants’ two-point changes were plotted on the same graph (i.e., visualizing older Blacks’ cognitive change as a group from ages 50 to 95). Understanding this age-graded pattern can improve the ways researchers approach age-related cognitive health issues among older Blacks and better adapt preventive efforts while offering a different perspective to disentangle the race–cognitive ability phenomenon. This study further hypothesized that this age-graded pattern would not follow a strict linear decline line due to the theory of selective survival.

According to the theory of selective survival (Dupre, 2007; Kim & Miech, 2009; Markides & Machalek, 1984), Black adults who have accumulated a lifetime of socioeconomic disadvantages and unfavorable health behaviors will experience disability and mortality at a younger age (Lariscy, 2017). Those who survive to advanced old age are likely to be a selective group of robust, healthier, and possibly also higher-educated and higher-socioeconomic status (SES) individuals. Although selective survival is often proffered as an explanation for the narrowing gaps in mortality and health disparities between Blacks and Whites with advancing age, the theory necessarily implies an aggregate change within older Blacks, given the healthier surviving population. When the survival effect begins to manifest, the impact of additional cumulative disadvantages diminishes (Kim & Miech, 2009), instantiating the age-as-leveler hypothesis in comparison between Blacks and Whites. Thus, looking solely within Blacks, it is reasonable to expect that as a group, their cognitive change may show a substantive difference around/after certain advanced age as the survival effect begins to manifest in cognitive health (Zahodne et al., 2016) and as additional disadvantages start to wane in influence. That is, when frail/disadvantaged individuals become unavailable for study, we may see the cognitive health and changes of oldest Blacks to be different—specifically, the slope will be less steep, showing a nonlinear age-graded pattern. In addition, based on previous studies (Kim & Miech, 2009; Lariscy, 2017; Markides & Machalek, 1984), it is likely that the survival effect for health occurs earlier (possibly in mid-70) than for mortality (~85).

Based on this central framework guiding the analyses of this study, we explored whether this nonlinear age pattern holds across four cognitive domains/abilities—inductive reasoning, declarative and working memory, and perceptual speed. These measures were chosen for several reasons. First, reasoning, memory, and speed are domains of cognitive ability considered to be sensitive to aging (Salthouse, 2010). Additionally, previously published studies using the BSBA data (e.g., Allaire et al., 2009) had used these cognitive domains as part of their criteria for mild cognitive impairment classification. Second, to facilitate ease of comparison across studies as these are commonly explored domains and commonly used measures in cognitive aging literature. For example, in Tucker-Drob and colleagues’ (2019) meta-analysis of the cognitive change literature, the domains derived for evaluation included speed, memory, and reasoning. Third, to ensure a more comprehensive coverage as variability across cognitive domains/tasks was also evidenced in the literature (Hultsch et al., 2002; Mella et al., 2018).

In sum, this study utilized data from the BSBA–Patterns of Cognitive Aging (BSBA–PCA), which was initiated to examine changes in the relationships between health, psychosocial factors, and cognition in Blacks across two waves at approximately 3 years (33 months) apart. We hypothesized that participants’ change/decline in cognitive performance from Wave 1 to 2 across four domains/abilities (inductive reasoning, declarative and working memory, and perceptual speed) is contingent upon the age at which they entered our study for testing, hence, attempted to characterize an age-graded pattern of change/decline visualizing older Blacks’ cognitive change as a group from ages 50 to 95. We believed that this age-graded pattern would not be linear but rather, there would be a substantive difference in the linear decline between Waves 1 and 2 around/after age 70–75 due to the selective survival effect. That is, the change/decline rate for young-old individuals would be faster but the change/decline rate for individuals after age 75 would be less steep when the effect begins to manifest. BSBA–PCA data allowed us to account for the unique effects of stress, financial strain, social support, and cardiovascular diseases risk that are often experienced by Blacks (Hummer et al., 2004). Therefore, perceived stress, financial strain, social support received and given, and cardiovascular risk were included in the analyses, together with sex, as covariates. Furthermore, in light of the evidence (Aiken-Morgan et al., 2018) that years of education have an effect on the global cognitive stability in Blacks, education was also included as a covariate.

Method

Participants

At Wave 1, the BSBA–PCA had 602 participants (74.58% female, n = 449), reported ages ranged from 48 to 95 (M = 69.12; SD = 9.76). At Wave 2, there were 450 participants (77.3% female, n = 348), reported ages ranged from 51 to 96 (M = 71.43, SD = 9.21). One hundred and fifty-two individuals were excluded after Wave 1 due to death (n = 58), unable to be found (n = 54), moved out beyond recruitment area (n = 21), too sick to participate (n = 13), and refusal (n = 6). T test analyses showed that individuals who were excluded were not significantly different from those who stayed in terms of sex, age, years and quality of education, financial strain (measured by income needs), perceived stress, working memory score, and social support received. However, they had a lower family income (t(576) = −2.09, p = .037), lower baseline Mini-Mental State Examination scores (t(600) = −2.23, p = .026), inductive reasoning (t(423) = −2.61, p = .009), declarative memory (t(575) = −2.74, p = .006), and perceptual speed (t(583) = −2.80, p = .005), showed higher scores for cardiovascular risk (t(584) = 2.06, p = .039), and provided less social support to friends and family (t(600) = −3.59, p < .001). For more details of the study procedure, refer to Aiken-Morgan and colleagues (2018).

Measures

Cognitive measures

The four cognitive domains were defined following the procedure outlined by prior published studies using the BSBA–PCA data (Allaire et al., 2009; Gamaldo et al., 2010). Inductive reasoning tested participants’ logical ability to identify related items. It required abstract reasoning and problem-solving skills, and was assessed using the Letter Series Test and Shipley Institute of Living Scale Abstraction Test. Declarative memory measured participants’ semantic processing ability and memory, and was assessed using the Hopkins Verbal Learning Task, Rey Auditory Verbal Learning Task, and Immediate Recall Test. Working memory required attention and short-term memory as it measured participants’ ability to retain and manipulate information. It was assessed using the Alpha Span Task, Operation Span Task, and Backward Digit Span task. Perceptual speed was assessed using the Number Comparison Test, Identical Pictures Test, and Digit Symbol Substitution Test, which tested participants’ ability to accurately and logically compare and react to stimuli. All the tests were assessed as total scores and all test scores were standardized to a mean of 50 and an SD of 10. Composite scores for the cognitive domains were computed by taking the sum of all the standardized test scores assessing that domain. High scores represent high cognitive performance.

Education and Covariates

Number of years of education was assessed. Sex was coded as 0 = male and 1 = female. The Perceived Stress Scale (Cohen et al., 1983) was used to measure the global feelings and thoughts about one’s situation in life during the past month. It has 14 items and participants were asked to appraise on a 5-point scale (0 = never to 4 = very often) how often they perceived situations such as “unable to control important things,” “unable to control irritations,” and “felt difficulties were piling up so high and could not be overcome” happened. Total possible score ranged from 0 to 48. Participants were asked to rate how well their income covered needs (financial strain) on a 4-point scale (0 = not very well, 1 = poorly but I get by, 2 = pretty well, and 3 = very well). Social support was measured by five items asking participants how frequently (0 = never to 3 = often) they provided/received household services, financial assistance, transportation, and companionship and advice to/from friends and family. Total possible score ranged from 0 to 15. Cardiovascular risk factor (CVRFS) was measured as a sum of self-reported number of diagnoses including diabetes, cardiovascular disease, high blood pressure, stroke, heart attack, angina, and circulation problems. The CVRFS composite score ranged from 0 = no/low risk to 7 = high risk. Covariates were assessed at both waves and the averages were used for analyses.

Statistical Analyses

Participants’ Wave 1 and 2 ages in year (to the nearest two decimals) on testing dates were calculated. Average age during the study period was used as the age variable for analyses. For the present study, six participants were excluded due to errors in/missing date of birth information, leaving n = 444 (76.1% female, n = 338), measured ages ranged from 51.5 to 95.5 years old (M = 70.1; SD = 9.2) included for analyses. Participants’ demographic, health, and cognitive information can be found in Table 1.

Table 1.

Demographic, Health, and Cognitive Characteristics of the Samples (n = 444)

Frequency Range Mean SD
Average age, yearsa 51.5–95.5 70.7 9.2
 >50–55 31
 >55–65 39
 >65–70 43
 >70–75 89
 >75–80 87
 >80–85 83
 >85–90 51
 >95 5
Education 3–20 11.7 2.9
Average perceived stress 3.5–44.5 18.7 7.4
Average financial strain 0–3 1.4 0.7
Average support received 0–14.5 7.6 3.0
Average support given 0–15 7.6 2.9
Average cardiovascular risk 0–6 2.1 1.4
Wave 1 cognitive composites
 Inductive reasoning 74–170 103.2 18.7
 Declarative memory 96–224 152.1 23.1
 Working memory 110–271 151.7 20.9
 Perceptual speed 88–245 151.9 24.9
Wave 2 cognitive composites
 Inductive reasoning 67–178 100.7 18.2
 Declarative memory 100–252 150.1 22.1
 Working memory 89–212 150.1 20.5
 Perceptual speed 97–222 150.2 23.3
Open in a new tab

Notes: Average = average across Waves 1 and 2.

aAge in fraction of a year calculated as the difference between testing/survey date and date of birth.

The data have a hierarchical structure in which the repeated measures are nested within individuals. We hypothesized that at least for some domains/abilities, Black adults’ slope of cognitive change/decline is contingent upon the age at which they entered our study for testing. This contingency should yield an age-graded pattern slopes from the age of 50 to 95 such that across this age distribution, the trend is nonlinear or does not follow an accelerated declining pattern. To test this, a multilevel model (MLM), allowing for simultaneous modeling of the within-person (level 1) two-wave effect, between-person (level 2) individual differences, and cross-level interaction (age-effect), was fitted separately with each of the inductive reasoning, declarative memory, working memory and perceptual speed composite score as outcome variable. The effect of age on the Wave 1 to Wave 2 change in outcome variables was tested in each model. A significant positive effect of age on the change means that the older a person was, the less they declined between the first and second time they were tested. Such a result indicates a nonlinear age-graded pattern across the entire age distribution.

Following the procedures outlined by Singer and Willett (2003), PROC MIXED on SAS 9.4 (SAS Institute, 2013) with full information maximum likelihood was used to generate parameter estimates and significance scores for all analyses. A variable for wave of measurement (0 = Wave 1 and 1 = Wave 2) was created to assess changes in outcome variables within-person across study time. Average age during the study was grand-mean centered at 70.71, and treated as a person-level variable, along with years of education at Wave 0 and sex. All other covariates (perceived stress, financial strain, support received/given, and cardiovascular risk) were grand-mean centered and entered as within-person variables. Mixed (fixed and random) effect unconditional means models were first fitted to partition the outcome variances and examine whether there were sufficient clustering and between-person changes to be explained by additional predictors. Intraclass correlation coefficients (ICCs) were computed using level 1 and level 2 variance parameters from the MLMs. Then, mixed-effect unconditional growth models were fitted with wave as the time predictor. This allowed for estimation of within-person changes across the study time points. All covariates were simultaneously added to this unconditional growth model and a cross-level interaction (between average study age and wave) was tested to see if the rate of change was nonuniform (i.e., nonlinear) across the age distribution.

Results

ICCs of .622 for declarative memory, .587 for working memory, and .735 for perceptual speed indicate fair amounts of clustering of Waves 1 and 2 scores within-person for these cognitive abilities, and much of the variabilities in scores (~59%–74%) were due to between-person differences (see Table 2 for parameter estimates of the unconditional means models). A moderately low ICC of .305 indicates less clustering of Wave 1 and 2 scores within-person for inductive reasoning. The inductive reasoning scales used in this study are well-established measures that have high (~.90) test–retest reliability (Kaya et al., 2012; Kyllonen et al., 2019). Thus, despite this lower ICC that suggests high within-person variability, a prior study (Allaire & Marsiske, 2005) has argued that within-person variability might be a consistent and reliable feature within a particular measure rather than an indication of random error or noise. In addition, literature has documented an ICC of .36 for inductive reasoning (Dzierzewski, 2007), which puts our moderately low ICC in alignment with previous work.

Table 2.

Unconditional Means Models

Inductive reasoning Declarative memory Working memory Perceptual speed
Coefficient (SE) Coefficient (SE) Coefficient (SE) Coefficient (SE)
Fixed effects
Initial status
 Intercept 101.49*** (0.76) 150.77*** (0.97) 150.60*** (0.86) 150.81*** (1.07)
Variance components
Within-person 236.81*** (19.19) 192.79*** (13.49) 251.20*** (18.94) 154.44*** (10.60)
Between-person
 Initial level/intercept 103.05*** (20.35) 316.69*** (28.87) 176.43*** (24.15) 427.78*** (34.62)
Open in a new tab

Note: ***p < .001.

The random intercept only Age as Moderator models indicate that age (on its own) had a significant negative main effect on how individuals performed over the study time (see Table 3). For every additional year in average age in the study, an average person’s cognitive score in each domain/ability decreased by between 0.27 and 0.81 points (specifically: inductive reasoning: −0.44, SE = 0.12, p < .001; declarative memory: −0.27, SE = 0.12, p = .029; working memory: −0.28, SE = 0.12, p = .020; perceptual speed: −0.81, SE = 0.11, p < .001). The time predictor, wave, did not significantly characterize changes in inductive reasoning, declarative memory, working memory, and perceptual speed abilities, possibly because of the hypothesized interaction between age and wave. Indeed, for inductive reasoning, there was a significant positive interaction between age and wave (0.34, SE = 0.14, p = .015), suggesting that older individuals in this study tended to have less negative change in inductive reasoning ability across an approximate 3-year period. Using predicted scores generated from the Age as Moderator model for inductive reasoning, Figure 1 visualizes this age-graded trend across the entire age distribution, indicating a leveling of decline at approximately 70–75 years old.

Table 3.

Age as Moderator Models

Inductive reasoning Declarative memory Working memory Perceptual speed
Coefficient (SE) Coefficient (SE) Coefficient (SE) Coefficient (SE)
Fixed effects
Initial status
 Intercept 100.43*** (1.67) 145.05*** (1.97) 146.71*** (1.72) 144.46*** (1.86)
 Age −0.44*** (0.12) −0.27* (0.12) −0.28* (0.12) −0.81*** (0.11)
 Sex 1.25 (1.72) 8.57*** (2.21) 5.26** (1.84) 9.14*** (2.08)
 Education 1.92*** (0.27) 1.67*** (0.34) 2.16*** (0.30) 2.76*** (0.32)
 Perceived stress −0.07 (0.09) −0.25** (0.09) −0.02 (0.09) −0.30*** (0.08)
 Financial strain 0.22 (0.83) −0.44 (0.87) 0.38 (0.87) 0.22 (0.81)
 Support received −0.54* (0.21) −0.50* (0.23) −0.53* (0.22) −0.57** (0.20)
 Support given 0.30 (0.23) 0.25 (0.25) 0.60* (0.24) 0.62** (0.22)
 Cardiovascular risk 0.25 (0.52) −0.54 (0.61) −1.10* (0.55) −1.53** (0.58)
Slope
 Intercept/wave −0.89 (1.26) −1.15 (1.04) 0.10 (1.27) −1.44 (0.94)
 Age (interaction) 0.34* (0.14) −0.17 (0.11) 0.03 (0.14) 0.15 (0.10)
Variance components
Within-person 240.15*** (20.63) 184.79*** (14.70) 269.91*** (21.74) 159.81*** (12.44)
Between-person
 Initial level/intercept 53.11** (18.50) 263.65*** (27.39) 86.64*** (20.82) 240.35*** (24.45)
Open in a new tab

Note: *p < .05. **p < .01. ***p < .001.

Figure 1.

Figure 1.

Open in a new tab

Visualizing change in inductive reasoning. Note: Change in inductive reasoning across study time by age (average age in the study as moderator) shows an age-graded trend, indicating a leveling of decline at approximately 70–75 years old.

To further explore the effect of the approximately 3-year study time period on change in inductive reasoning ability conditioned on different average ages, the region of significance was estimated using the Plot of Conditional Effects and Johnson–Neyman Technique in R (Bachl, 2015; Figure 2). The conditional effect suggests that for Black adults ~64 years and younger, the change in inductive reasoning ability across study time was significantly negative, whereas for those older than 63.71, the change was nonsignificant.

Figure 2.

Figure 2.

Open in a new tab

Conditional effect of study time on change in inductive reasoning. Note: Conditional effect of the time predictor (wave) as a function of average age during study time. Gray area denotes the confidence intervals. Significance region is denoted by area where the confidence intervals do not cross zero.

Discussion

Longitudinal studies have shown that cognitive decline in older adulthood is inevitable and becomes faster with increasing age. Studies using data from the Swedish Adoption/Twin Study of Aging (e.g., Finkel et al., 2007) and the Seattle Longitudinal Study (SLS; e.g., Gerstorf et al., 2011) demonstrate that older cohorts decline faster than younger in many cognitive domains/abilities. However, given the literature in cognitive aging and specifically these two longitudinal data sets, which captured predominately White participants, this conclusion is not necessarily generalizable to other racial groups. Literature looking at the disparity between Blacks and Whites suggests that during late adulthood, differences in mortality rates and functional and health disadvantages disappear (Kim & Miech, 2009; Markides & Machalek, 1984). The convergence point for mortality rates is said to be around 85 and for health and functional disadvantages is around 71. This leveling phenomenon is supported by the selective survival theory (Markides & Machalek, 1984), which implies that Blacks who survive past an advancing age are healthier and more robust (Gibson, 1994). Thus, it is very likely that older Blacks’ cognitive health does not follow the accelerated decline trend found in general older White population across tasks of fluid intelligence.

This study looked only within Black adults to examine if, as a group, their cognitive change across adulthood (in various domains) shows a unique age-graded pattern rather than a linear or accelerated decline across the age distribution. As hypothesized, cognitive change over an approximately 3-year period assessed by inductive reasoning ability showed age-graded differences in the rate of decline. After accounting for the effects of education and health and social stressors/supports unique to Blacks, advancing age actually moderated the negative effect of the 3-year time period on inductive reasoning ability such that around 70–75, the decline leveled out and around 80–85, there was a slight general upward tendency. This is akin to and consistent with Kim and Miech’s (2009) study looking at the age trajectories difference in functional health between Blacks and Whites. They found that after controlling for SES, the effect of cumulative disadvantages still exits among young and middle-aged Black adults—that is, their slope of health change remained steep until after ~71, Blacks’ slope of health change became less steep.

However, we did not find significant age-graded differences in cognitive ability assessed by declarative memory, working memory, or perceptual speed. These findings did not complement the literature that suggests a single general factor of cognitive ability (see Tucker-Drob et al., 2019). Tucker-Drob and colleagues’ (2019) meta-analysis indicates that studies found significant correlations in longitudinal changes among cognitive abilities, suggesting age-related cognitive changes are likely to occur in tandem such that individuals who show decline in one domain are likely to show decline in another domain as well. The reasons for our contrasting results can elude simple answers, as Salthouse (2010) pointed out that we may never have a definitive answer as to the number of different types of age-related cognitive and neuropsychological variables influencing changes. Nonetheless, we speculate the following for why age-graded pattern was only found for inductive reasoning in this study. First, it is possible that some crystallized intelligent abilities (e.g., experience and logical strategy which tend to improve with age) were used to facilitate the completion of reasoning tasks such that participants in this study showed a slower rate of decline in inductive reasoning with advanced age. This might be particularly salient within the Black population because they often acquire unidentified learning strategies from life experiences that common neuropsychological testing does not assess (Whitfield & Aiken-Morgan, 2008). Second, it is also possible that early-life education attainment played a role in older individuals’ reasoning ability. Studies (Gerstorf et al., 2011; Schaie et al., 2005) have suggested that educational processes and structures at different historical times may have bestowed individuals with different training and analytic ability that is subsequently reflected in cohort differences in psychometric tests. Whitfield and colleagues (2010) also found that education can be a source of variability across fluid mechanic and crystallized-pragmatic abilities among Black adults. In addition, some consider education as reflective of life experience beyond age and this often serves as a proxy for cognitive reserves that protect against manifestation of brain disease (Siedlecki et al., 2009). Third, the results of survival effect might simply manifest more strongly on inductive reasoning because of the required problem-solving skill that often builds on life experience. Notwithstanding, studies not supporting a single general factor of cognitive ability are not unprecedented in the literature. For example, Finkel and colleagues’ (2007) Swedish Adoption/Twin Study of Aging did not find a significant difference between younger and older age cohorts in the rate of decline for processing speed ability despite finding significant cohort differences for several other abilities.

Gerstorf and colleagues’ (2011) study looking at cohort differences in the rate of cognitive aging found typical linear decline with acceleration in five cognitive domains: spatial orientation, word fluency, verbal meaning, number ability, and inductive reasoning. One reason this study found the opposite pattern for the rate of change/decline for inductive reasoning may be that, as emphasized throughout, the age-graded pattern found seems to be unique to older Black adults. Gerstorf and colleagues’ (2011) study utilized data from the SLS, the racial breakdown of which was not clearly stated. Using a population sample like the SLS can produce nationally representative results, but because the Black sample size may be relatively small, it could easily mask the unique effect specific to the population of older Black adults.

Taken together, there are clinical and preventive implications for Blacks’ cognitive health if as a racial group they do not follow the general linear or accelerated cognitive age-graded pattern. Common cognitive intervention programs generally target older adults 65+ years old, but if decline is most significant among younger and middle-aged Blacks, these help and prevention efforts may come too late. Early interventions for cognitive maintenance or improvement may be more relevant and urgently needed for Black adults. Besides maintenance and preventive efforts, equally important is the effort to improve early life and equal educational opportunity and quality for Blacks. Among the covariates examined in this study, education appears to be a clear protective factor against cognitive decline. Prior research has also demonstrated the well-established and cumulative association between educational attainment in early life and better health-related outcomes in later life (Kye et al., 2014), and education as a proxy for cognitive reserve (Siedlecki et al., 2009). Irrefutably, societal and institution efforts to push for equitable educational resources for Black youth are much needed (Orfield & Lee, 2005). Furthermore, this negative decelerating age-graded pattern of change/decline in Black adults also raises the question of whether there are specific sociodemographic or psychological characteristics associated with those oldest Blacks who defy the negative effect of chronological age on cognitive health. These characteristics could be protective factors worth further exploration, and could also have important implications for improving the cognitive health of other racial groups. Nonetheless, future studies can explore other characteristics such as personality traits, family and work environment, and social network, etc., that might be associated with cognitive health among the oldest Blacks.

One limitation of this current study is that there are only two time points per individual, which disallowed direct assessment of within-person longitudinal trajectories. If more waves of data were available, individuals’ growth trajectories could be estimated and an age convergence test could be implemented to verify if individuals’ longitudinal growth merges onto a common trajectory. Another limitation is that this study did not find the unique age-graded pattern in three (declarative memory, working memory, and perceptual speed) of the four cognitive abilities tested. Although this may simply be a result of limitation in the power to identify the age-graded pattern from only two time points per person, it certainly limits our ability to rule out the possibility of a chance finding and conclude for certain that this represents specific differences among different cognitive domains. Future research with more measurements for each individual may be needed to explore differences in decline in these cognitive domains for older Blacks.

In sum, findings from this study suggest that among older Blacks, the decline of inductive reasoning ability slows down after/around age 70. This might shift the focus of cognitive intervention programs as early interventions for cognitive maintenance or improvement may be more urgently needed for Blacks.

Acknowledgments

Special thanks to the BSBS-PCA staff for data collection and data entry. The data will be made available upon request. The study reported in this manuscript was not preregistered.

Funding

This work was supported by the National Institute on Aging (R01 AG24108 and AG24108-S1 to K. E. Whitfield and J. C. Allaire; 02AG059140 and P30AG059298 to R. J. Thorpe Jr.; and R01 AG054363 to R. J. Thorpe Jr. and K. E. Whitfield) and the National Institute on Minority Health and Health Disparities (U54MD000214 to R. J. Thorpe Jr.).

Conflict of Interest

None declared.

References

  1. Aiken-Morgan, A. T., Gamaldo, A. A., Wright, R. S., Allaire, J. C., & Whitfield, K. E. (2018). Stability and change in cognitive status classification of black older adults. Journal of the American Geriatrics Society, 66(1), 179–183. doi: 10.1111/jgs.15225 [DOI] [PubMed] [Google Scholar]
  2. Aiken-Morgan, A. T., Marsiske, M., Dzierzewski, J. M., Jones, R. N., Whitfield, K. E., Johnson, K. E., & Cresci, M. K. (2010). Race-related cognitive test bias in the active study: A mimic model approach. Experimental Aging Research, 36(4), 426–452. doi: 10.1080/0361073X.2010.507427 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Aiken-Morgan, A. A., Marsiske, M., & Whitfield, K. E. (2008). Characterizing and explaining differences in cognitive test performance between African American and European American older adults. Experimental Aging Research, 34(1), 80–100. doi: 10.1080/03610730701776427 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Allaire, J. C., Gamaldo, A., Ayotte, B. J., Sims, R., & Whitfield, K. (2009). Mild cognitive impairment and objective instrumental everyday functioning: The everyday cognition battery memory test. Journal of the American Geriatrics Society, 57(1), 120–125. doi: 10.1111/j.1532-5415.2008.02054.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Allaire, J. C., & Marsiske, M. (2005). Intraindividual variability may not always indicate vulnerability in elders’ cognitive performance. Psychology and Aging, 20(3), 390–401. doi: 10.1037/0882-7974.20.3.390 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Alley, D., Suthers, K., & Crimmins, E. (2007). Education and cognitive decline in older Americans: Results from the AHEAD sample. Research on Aging, 29(1), 73–94. doi: 10.1177/0164027506294245 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bachl, M. (2015, March). Plot of conditional effects and Johnson-Neyman technique in R. RPubs by RStudio. https://rpubs.com/bachl/jn-plot [Google Scholar]
  8. Bénabou, R. (1994). Human capital, inequality, and growth: A local perspective. European Economic Review 38(3–4), 817–826. doi: 10.1016/0014-2921(94)90118-X [DOI] [Google Scholar]
  9. Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24(4), 385–396. doi: 10.2307/2136404 [DOI] [PubMed] [Google Scholar]
  10. Dupre, M. E. (2007). Educational differences in age-related patterns of disease: Reconsidering the cumulative disadvantage and age-as-leveler hypotheses. Journal of Health and Social Behavior, 48(1), 1–15. doi: 10.1177/002214650704800101 [DOI] [PubMed] [Google Scholar]
  11. Dzierzewski, J. (2007). Intraindividual variability in objective and subjective sleep and cognitive performance in older adults with insomnia [Master’s thesis]. The University of Florida. http://etd.fcla.edu/UF/UFE0020955/dzierzewski_j.pdf [Google Scholar]
  12. Evans, M. K., Rosenbaum, L., Malina, D., Morrissey, S., & Rubin, E. J. (2020). Diagnosing and treating system racism. New England Journal of Medicine, 382, e111–113. doi: 10.1056/NEJMe2021693 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Finkel, D., Reynolds, C. A., McArdle, J. J., & Pedersen, N. L. (2007). Cohort differences in trajectories of cognitive aging. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 62(5), P286–P294. doi: 10.1093/geronb/62.5.p286 [DOI] [PubMed] [Google Scholar]
  14. Gamaldo, A. A., Allaire, J. C., Sims, R. C., & Whitfield, K. E. (2010). Assessing mild cognitive impairment among older African Americans. International Journal of Geriatric Psychiatry, 25(7), 748–755. doi: 10.1002/gps.2417 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gerstorf, D., Ram, N., Hoppmann, C., Willis, S. L., & Schaie, K. W. (2011). Cohort differences in cognitive aging and terminal decline in the Seattle Longitudinal Study. Developmental Psychology, 47(4), 1026–1041. doi: 10.1037/a0023426 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Gibson, R. C. (1994). The age-by-race gap in health and mortality in the older population: A social science research agenda. The Gerontologist, 34(4), 454–462. doi: 10.1093/geront/34.4.454 [DOI] [PubMed] [Google Scholar]
  17. Goh, J. O., An, Y., & Resnick, S. M. (2012). Differential trajectories of age-related changes in components of executive and memory processes. Psychology and Aging, 27(3), 707–719. doi: 10.1037/a0026715 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Grady, M., & Edgar, T. (2003). Appendix D: Racial disparities in healthcare: Highlights from focus group findings. In Smedley B. D., Stith A. Y., & Nelson A. R. (Eds.), Unequal treatment: Confronting racial and ethnic disparities in health care (pp. 392–405). National Academies Press. [PubMed] [Google Scholar]
  19. Hultsch, D. F., MacDonald, S. W., & Dixon, R. A. (2002). Variability in reaction time performance of younger and older adults. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 57(2), P101–P115. doi: 10.1093/geronb/57.2.p101 [DOI] [PubMed] [Google Scholar]
  20. Hummer, R., Maureen, B., & Rogers, R. (2004). Racial and ethnic disparities in health and mortality among the U.S. elderly population. In Anderson N., Bulatao R., & Cohen B. (Eds.), Critical perspectives on racial and ethnic differences in health in late life (pp. 53–94). National Academies Press. [PubMed] [Google Scholar]
  21. Kaya, F., Delen, E., & Bulut, O. (2012). Test review: Shipley-2 manual. Journal of Psychoeducational Assessment, 30(6), 593–597. doi: 10.1177/0734282912440852. [DOI] [Google Scholar]
  22. Kim, J., & Miech, R. (2009). The Black–White difference in age trajectories of functional health over the life course. Social Science & Medicine (1982), 68(4), 717–725. doi: 10.1016/j.socscimed.2008.12.021 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Kye, B., Arenas, E., Teruel, G., & Rubalcava, L. (2014). Education, elderly health, and differential population aging in South Korea: A demographic approach. Demographic Research, 30(26), 753–794. doi: 10.4054/DemRes.2014.30.26. [DOI] [Google Scholar]
  24. Kyllonen, P., Hartman, R., Sprenger, A., Weeks, J., Bertling, M., McGrew, K., Kriz, S., Bertling, J., Fife, J., & Stankov, L. (2019). General fluid/inductive reasoning battery for a high-ability population. Behavior Research Methods, 51(2), 507–522. doi: 10.3758/s13428-018-1098-4 [DOI] [PubMed] [Google Scholar]
  25. Lariscy, J. T. (2017). Black-white disparities in adult mortality: Implications of differential record linkage for understanding the mortality crossover. Population Research and Policy Review, 36(1), 137–156. doi: 10.1007/s11113-016-9415-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. LaVeist, T. A., Gaskin, D., & Richard, P. (2011). Estimating the economic burden of racial health inequalities in the United States. International Journal of Health Services: Planning, Administration, Evaluation, 41(2), 231–238. doi: 10.2190/HS.41.2.c [DOI] [PubMed] [Google Scholar]
  27. Lee, H. B., Richardson, A. K., Black, B. S., Shore, A. D., Kasper, J. D., & Rabins, P. V. (2012). Race and cognitive decline among community-dwelling elders with mild cognitive impairment: Findings from the Memory and Medical Care Study. Aging & Mental Health, 16(3), 372–377. doi: 10.1080/13607863.2011.609533 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Manly, J., & Mayeux, R. (2004). Ethnic differences in dementia and Alzheimer’s disease. In Anderson N., Bulatao R., & Cohen B. (Eds.), Critical perspectives on racial and ethnic differences in health in late life (pp. 95–141). National Academies Press. [PubMed] [Google Scholar]
  29. Markides, K. S., & Machalek, R. (1984). Selective survival, aging and society. Archives of Gerontology and Geriatrics, 3(3), 207–222. doi: 10.1016/0167-4943(84)90022-0 [DOI] [PubMed] [Google Scholar]
  30. Mella, N., Fagot, D., Renaud, O., Kliegel, M., & de Ribaupierre, A. (2018). Individual differences in developmental change: Quantifying the amplitude and heterogeneity in cognitive change across old age. Journal of Intelligence, 6(1), 1–10. doi: 10.3390/jintelligence6010010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Orfield, G. & Lee, C. (2005, January). Why segregation matters: Poverty and educational inequality. Civil Rights Project, Harvard University. [Google Scholar]
  32. Salthouse, T. A. (2010). Selective review of cognitive aging. Journal of the International Neuropsychological Society, 16(5), 754–760. doi: 10.1017/S1355617710000706 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. SAS Institute . (2013). SAS 9.4 software. SAS Institute Inc. [Google Scholar]
  34. Schaie, K. W., Willis, S. L., & Pennak, S. (2005). An historical framework for cohort differences in intelligence. Research in Human Development, 2(1–2), 43–67. doi: 10.1080/15427609.2005.9683344 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Siedlecki, K. L., Stern, Y., Reuben, A., Sacco, R. L., Elkind, M. S. V., & Wright, C. B. (2009). Construct validity of cognitive reserve in a multi-ethnic cohort: The Northern Manhattan study. Journal of the International Neuropsychological Society, 15(4), 558–569. doi: 10.1017/S1355617709090857. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Singer, J., & Willett, J. (2003). Applied longitudinal data analysis: Modeling change and event occurrence. Oxford University Press. [Google Scholar]
  37. Tucker-Drob, E. M., Brandmaier, A. M., & Lindenberger, U. (2019). Coupled cognitive changes in adulthood: A meta-analysis. Psychological Bulletin, 145(3), 273–301. doi: 10.1037/bul0000179 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Weuve, J., Barnes, L. L., Mendes de Leon, C., Rajan, K. B., Beck, T., Aggarwal, N. T., Hebert, L. E., Bennett, D. A., Wilson, R. S., & Evans, D. A. (2018). Cognitive aging in Black and White Americans: Cognitive, cognitive decline, and incidence of Alzheimer disease dementia. Epidemiology, 29(1), 151–159. doi: 10.1097/EDE.0000000000000747 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Whitfield, K. E., & Aiken Morgan, A. T. (2008). Minority populations and cognitive aging. In Hofer S. M. & Alwin D. F. (Eds.). Handbook of cognitive aging: Interdisciplinary perspectives (pp. 384–397). Sage Publications. [Google Scholar]
  40. Whitfield, K. E., Allaire, J. C., Belue, R., & Edwards, C. L. (2008). Are comparisons the answer to understanding behavioral aspects of aging in racial and ethnic groups? The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 63(5), P301–P308. doi: 10.1093/geronb/63.5.p301 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Whitfield, K. E., Allaire, J. C., Gamaldo, A. A., & Bichsel, J. (2010). Factor structure of cognitive ability measures in older African Americans. Journal of Cross-Cultural Gerontology, 25(3), 271–284. doi: 10.1007/s10823-010-9120-z [DOI] [PubMed] [Google Scholar]
  42. Wilson, R., Beckett, L., Barnes, L., Schneider, J., Bach, J., Evans, D., & Bennett, D. (2002). Individual differences in rates of change in cognitive abilities of older persons. Psychology and Aging, 2, 179–193. doi: 10.1037//0882-7974.17.2.179 [DOI] [PubMed] [Google Scholar]
  43. Zahodne, L. B., Manly, J. J., Azar, M., Brickman, A. M., & Glymour, M. M. (2016). Racial disparities in cognitive performance in mid- and late adulthood: Analyses of two cohort studies. Journal of the American Geriatrics Society, 64(5), 959–964. doi: 10.1111/jgs.14113 [DOI] [PMC free article] [PubMed] [Google Scholar]

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

RESOURCES