The Long-Term Impact of Racial and Ethnic School Composition on Late-Life Cognition, Depressive Symptoms, and Overall Health Among Older US Latinos

Abstract

Latinx individuals’ educational experiences in the US were shaped by structural and interpersonal discrimination, potentially contributing to disparities in dementia and related risk factors. Racial and ethnic school composition reflects resources and co-ethnic belonging.

We used Health and Retirement Study data from 315 Latinx adults aged≥50 years who reported their school racial/ethnic composition through 12^th grade. We estimated associations between each composition type (1-year unit) and memory scores (0–20), depressive symptoms (0–8), and poor self-rated health (SRH; yes/no), adjusting for sociodemographic factors.

Greater duration in schools serving majority Latinx students was not associated with memory (β=−0.01 [−0.09, 0.07]) or depressive symptoms (β=−0.03 [−0.09, 0.04]) but was associated with higher risk for poor SRH (RR=1.05 [1.01, 1.08]). Greater duration in schools serving majority Black students was associated with lower memory (β=−0.16 [−0.31, −0.01]), more depressive symptoms (β=0.18 [0.07, 0.29]), and higher risk for poor SRH (RR=1.06 [1.00, 1.11]). Greater duration in schools serving majority White students was associated with higher memory scores (β=0.09 [0.01, 0.16]), lower depressive symptoms (β=−0.05 [−0.11, 0.00]), and decreased risk for poor SRH (RR=0.91 [0.87, 0.96]).

Among Latinx adults, the impact of school composition varied by whether schools served other marginalized populations, reflecting countervailing influences.

INTRODUCTION

Latino older adults (hereafter, Latino/Latina/Latine will be referred to as Latinx) are disproportionately affected by Alzheimer’s disease and related dementias,¹ as well as dementia risk factors.^2–4 This pattern may be partly influenced by the significant educational attainment gap between Latinx and other racial groups in the US.^5–7 Education plays a fundamental role in shaping various aspects of well-being that may contribute to dementia risk.^8–11 Historical educational experiences for Latinx older adults in the US have been shaped by structural racism, including discriminatory laws and covert practices that enforced segregation and led to the inequitable distribution of school funding and resources that disadvantaged racially marginalized students.¹² While emerging studies have examined how segregated schooling has impacted the cognitive health of older Black adults,^13–17 limited research has examined its impacts on older Latinx adults.

Historical Context on School Segregation Among Latinos

Historical schooling experiences for Latinx students have been heterogeneous across the US, shaped by a combination of racial, ethnic, and linguistic factors.^18,19 Education has served as a double-edged sword, acting as both a tool of oppression and as a mechanism for social control and forced assimilation–a practice that dates to the Spanish colonial period, when Indigenous populations were subjugated under the pretext of education (e.g., within California Missions).²⁰ In the early 20^th century, the economic uncertainty and civil unrest following World War I and the Great Depression gave rise to anti-immigrant sentiments and xenophobic policies to combat increasing immigration trends.²¹ These sentiments were formalized in laws like the 1924 Immigrant Act and reinforced by the subsequent mass deportations of Mexican-Americans.²¹ They were also perpetuated more subtly through informal practices that segregated public institutions (e.g., parks, pools, movie theaters),²² targeting Latinx communities by weaponizing language as an exclusionary tool to justify racial discrimination in schools and forced assimilation.²³

Education policies varied drastically across and within states, with more explicit language pertaining to the de jure segregation of Black students in legal statutes.¹² Given that school segregation policies were often decided at the local level, official documentation of the extent and geographic patterns of segregation is limited. In states like Texas and California, with historically greater Latinx populations and early settler displacement (e.g., the Manifest Destiny), historical records show that Latinx children were often separated from White children and placed in “Mexican” schools.^24–26 For example, by the 1930s, an estimated 85% of California schools and 90% of schools in South Texas practiced segregation of Latinx students at the classroom or school level.^25,27 While this practice was most pronounced in the Southwest, it extended beyond these regions across the US.²⁸

“Mexican” schools and classrooms were typically underfunded and under-resourced;²⁹ the justification for this rested on the frequently inaccurate assumption that Latinx children were learning English as a second language.²³ As a result, several lawsuits were filed challenging the segregation of Latinx students.²⁵ One landmark case was Mendez v. Westminster School District (1946), in which a California court ruled that segregation based on national origin or ancestry was unconstitutional, setting the precedent of desegregation in California and, by extension, the US.¹⁸ Similar cases in Texas, including Delgado v. Bastrop Independent School District (1948) and LULAC v. Richards (1950s), challenged discriminatory school segregation practices.²⁵ Given the binary framing of race in the 1954 Brown v. Board of Education Supreme Court ruling (i.e., segregation of Black students from White students) and the legal classification of Latinx populations as White, the desegregation of Latinx students in multiracial school districts with multiple marginalized groups remained legally and politically ambiguous well into the 1970s (e.g., Keyes v. School District Number One).^19,23,30 Despite efforts to promote desegregation, resistance remained strong, and economic and racial segregation in schools persists in the US.^31,32

Prior Research

There is limited research on the long-term health impacts of school segregation, or school racial and ethnic composition more generally, for Latinx older adults. Some studies have focused on more short-term educational, psychosocial, and behavior-related outcomes. For example, prior quasi-experimental studies have shown that school desegregation efforts following the Mendez court decision in California led to an average increase of 0.88 years in educational attainment (a relative increase of 8.6%) by adulthood.³³ One study found no significant health differences in depression scores, self-rated health, and alcohol abuse among Latinx adults (ages 18–32) who attended a school with a higher percentage of non-Hispanic White students compared to those who did not.³⁴ However, another study suggested that attending a school with a greater proportion of non-Hispanic White students was associated with an increased risk of smoking behavior in Latinx adults.³⁵ A systematic review found evidence that greater classroom diversity was associated with less loneliness and higher self-worth in Latinx students.³⁶ An emerging body of work examining the long-term health consequences of segregated schooling for older Black adults has found associations with lower cognitive function^13–15,17 and worse cardiovascular health outcomes,³⁷ but fewer depressive symptoms.^36,38 Such studies have not been extended to include older Latinx adults, likely due to the very limited available data on this topic.

Mechanisms Linking School Segregation and Dementia Risk

Several mechanisms have been proposed to explain how experiencing segregated schooling might influence late-life dementia risk. First, segregated schools have historically received fewer resources and funding, and often had shorter term lengths, which impacted both the quality and quantity of educational attainment.^39–43 Lower education quality and quantity is in turn linked to lower access to socioeconomic and health resources (e.g., material, informational) in mid-life.^41,44–47 Mid-life socioeconomic status is a critical factor in the prevention and management of chronic health conditions in later life.^42,48 Second, the inequitable conditions of segregated schools occur during childhood and adolescence, sensitive periods characterized by rapid neurodevelopment, such that they might play a significant role in shaping cognitive outcomes.^43,44,49,50

A third potential mechanism, that may mitigate the adverse impacts of fewer resources, is that segregated schools could have shielded racially marginalized students from potential interpersonal discrimination from peers and teachers that commonly occurred in desegregated schools.^51,52 Across the US, desegregation policies did not necessarily lead to integration, as local resistance meant many schools remained predominantly composed of specific racial or ethnic groups and lacked meaningful diversity and inclusion. Recent studies have linked student-teacher racial concordance and greater school diversity with students’ well-being through the mitigation of stress via emotional and instructional support, in addition to academic benefits.^53–56 However, much of this research has focused on Black populations, and Latinx adults may have experienced similar challenges when navigating racialized schooling environments. Latinx individuals come from a range of national backgrounds, and internalized biases, xenophobia, and colorism may shape how they experience discrimination in these settings.⁵⁷ As a result, it remains unclear to what extent these processes influence late-life health risks of older Latinx adults.

The Present Study

This study investigated associations between school racial and ethnic composition and 3 key factors linked to later-life dementia risk: verbal memory, depressive symptoms, and self-reported health among older Latinx adults who attended school in the US. These factors were chosen to identify key pathways linking school composition to dementia risk. We hypothesized that attending schools with a majority composition of marginalized students, such as Black or Latinx students, would be associated with lower verbal memory function and poorer general health, given the inequitable distribution of resources across schools by racial and ethnic composition. However, we hypothesized that Latinx students may benefit from a sense of identification or belonging with their peers and teachers in schools with a majority composition of Latinx students, which may mitigate some of the potential negative effects of attending a lower-resourced school on psychosocial well-being. Therefore, following prior studies finding protective associations between school segregation and depressive symptoms for older Black adults, we expected to see protective or null associations between attending a majority Latinx school for Latinx students.

METHODS

Study population

Data came from the Health and Retirement Study (HRS), a nationally representative cohort study of US community-dwelling adults over the age of 50. Initiated in 1992, the survey has been fielded approximately every 2 years with age-eligible respondents and their spouses. The HRS provides a rich resource of comprehensive health, psychosocial, and economic data across the participants’ lifecourse. Further description of the HRS study can be referenced elsewhere.⁵⁸

In 2015, the HRS collected information on respondents’ residential and schooling history through a Life History Mail Survey (LHMS). The abbreviated HRS-LHMS was initially mailed to a subgroup of 11,256 HRS respondents and their spouses who participated in the 2014 core interview. Just over half completed the survey (n=6481; 58%). In 2017, the LHMS was expanded and re-administered to a majority of newly eligible participants who received a full version or the supplemental version based on prior participation (n= 10,508; response rates: 58%–74%). A similar strategy was employed in 2019, during which 3,388 unique participants received the Fall or Spring LHMS survey (response rates: 24%–37%).⁵⁹ These analyses utilize the Harmonized 2015–2017 LHMS Cross-Wave file and Fall/Spring 2019 LHMS datasets available in May 2024.

We first restricted our analytic sample to 2015–2019 HRS-LHMS participants who self-identified as Latino or Hispanic (n=1,091). Given we were interested in the participants’ racialized schooling experiences within the US, we restricted our analytic sample to US-born participants or those who migrated to the US by the age of 17 (n=538). We further restricted our sample to participants with any school composition information (n=362; 33% of those excluded had not attended schooling in the US), at least one measure of verbal memory, depressive symptomatology, and self-rated health between 1995 and 2016 (n=361), and had complete covariate data (n=361). The final analytical sample included 315 respondents ages 50 years or older (Figure S1). Included participants were more likely to be US-born, have higher education, and have more educated parents (Table S1).

Measures

K-12 duration in schools of a given racial and ethnic composition

Participants self-reported each school they attended from kindergarten through 12th grade. For each school, respondents provided the first and last grade at that school, their ages at those corresponding grades, and the racial and ethnic composition of the majority of students at that school (used as a proxy for school segregation). Response options for the latter question were “1) White, 2) Black, 3) Hispanic/Latino, or 4) Other” (hereafter, another racial/ethnic composition). Participants could contribute time to each school composition type, which were modeled independently. From this we calculated the total number of years spent at a school serving a predominantly racially/ethnically concordant student body (hereafter, majority Latinx students); a racially/ethnically discordant student body that is not predominately comprised of a marginalized population (hereafter, majority White students); and a racially/ethnically discordant student body that is predominantly comprised of a marginalized population (hereafter, majority Black students or other marginalized groups), with values ranging from 0 to 13 years. A small proportion of participants attended a school serving predominantly Black students for at least one year (5%); we combined their values and those with the “Other” racial/ethnic composition type in the primary analysis. For non-US-born participants, only the data from the years or grades following their age of migration were included.

Outcomes

Verbal memory scores. The HRS utilizes a modified version of the Telephone Instrument for Cognitive Status (TICS) to assess cognitive function in both face-to-face interviews and by telephone. Our analysis focused on verbal memory due to its sensitivity to early cognitive changes and its relevance to Alzheimer’s disease risk.⁶⁰ Baseline memory scores were assessed through an immediate word-recall task, in which respondents were read a list of 10 common nouns and immediately asked to recall as many words as possible, followed by a delayed recall task, conducted approximately 5 minutes later, in which respondents were asked to recall the same 10 words. The final score was calculated as a sum of the scores of these two tasks (range: 0–20) with higher scores indicating better memory.

Depressive symptoms. Past-week depressive symptoms were assessed utilizing an eight-item modified Center for Epidemiologic Studies Depression Scale.⁶¹ Items were reported as binary “yes/no” responses, and the score was generated as the sum of six “negative” indicators plus the inverse of two “positive” indicators. The summary counts ranged from 0 to 8, with higher values indicating more depressive symptoms.

Poor general health. Participants were asked to rate their overall health using a 5-point Likert scale, ranging from 1 (Excellent) to 5 (Poor). Responses were then collapsed into a binary poor general health variable (“Excellent”, “Very good”, and “Good” (referent group) vs. “Poor” and “Fair” health).

Covariates

Confounders included baseline age, sex/gender, region of birth, a birth cohort indicator, and parental education. Details on the operationalization of confounders and the covariates included in sensitivity analyses (e.g., educational attainment, public school status) are provided in Appendix S1.

Statistical analysis

We calculated descriptive statistics for the overall sample. We then implemented multivariate linear regressions to separately estimate the association of duration of exposure to school racial and ethnic composition category on memory function and depressive symptoms, adjusting for demographics, parental education, birth region, and birth cohort. We estimated the association between duration of exposure to school racial and ethnic composition and poor self-rated health via modified Poisson regression with a log link and robust standard errors, controlling for previously listed covariates.

We performed several sensitivity analyses. First, despite inherent tradeoffs in sample size and variance, we estimated separate results using a school composition variable that distinguished between schools where the majority of students were Black and those where students were another racial/ethnic composition. Second, we additionally adjusted for educational attainment, conceptualized as a mediator. These analyses informally elucidate how much of the associations between schools’ racial and ethnic compositions and late-life outcomes are driven by differential socioeconomic resources. We also adjusted for public-school status to account for other schooling characteristics that could moderate or mediate these associations. Third, we restricted our sample to participants who attended the entirety of their schooling in the US to reduce additional confounding related to selective migration. Fourth, we included weights to account for LHMS participation (Appendix S2). Analyses were performed using Stata 18.0 (StataCorp, College Station, TX).

RESULTS

Participant characteristics are shown in Table 1 and Table S2. Participants in the analytic sample had a baseline average age of 55.7 years (4.5 ± SD), 87% were born in the US, of which 52% were born in a non-Southern state, 65% were women, and over 40% had parents who completed at least 8 years of education. On average, participants spent 3.7 years (4.5 ± SD) in schools serving a majority Latinx student body, 4.3 years (4.8 ± SD) in schools serving a majority White student body, and 0.8 years (2.3 ± SD) in schools serving majority Black or other marginalized students.

Table 1.

Baseline descriptive characteristics (mean [SD] or %) of the analytical sample of participants from the HRS-LHMS 2015–2019 (n=315)

	Overall
Respondent demographic characteristics
Age, years	55.7 (4.5)
Sex
Male	44.1%
Female	55.9%
Own educational attainment a	12.4 (3.2)
Mother’s educational attainment
0–7 years	41.9%
8+	48.3%
Missing	9.8%
Father’s educational attainment
0–7 years	39.1%
8+	43.8%
Missing	17.1%
Birthplace
US South b	34.6%
US Non-South	52.1%
Non-US born	13.3%
School type c
Only attended public schools	77.8%
Only attended private schools	3.2%
Attended both private and public schools	18.5%
Duration of time in schools based on racial and ethnic composition, in years (mean, SD)
Racially/ethnically concordant (majority Latinx)	3.7 (4.5)
Racially/ethnically discordant, non-marginalized (majority White)	4.3 (4.8)
Racially/ethnically discordant, marginalized (majority Black or another marginalized group)	0.8 (2.3)
Majority Black d	0.3 (1.3)
Attended at least one grade with the following racial and ethnic composition type (% yes)
Majority Latinx	55.6%
Majority White	56.2%
Majority Black or another marginalized group	14.9%
Majority Black d	5.1%

^a.3 participants were missing values for their educational attainment.

^b.Birth in the Southern United States defined using the U.S. Census Bureau regional designations, this included the following states: Alabama, Arkansas, Delaware, District of Columbia, Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia.

^c.2 participants were missing public vs. private school designation.

^d.Descriptives statistics provided for alternative variable specification utilized in sensitivity analyses.

School composition and verbal memory

Duration of attendance in schools serving a majority Latinx student body was not associated with verbal memory scores (b [95% CI]= −0.01 [−0.09, 0.07]). In contrast, each additional year spent in schools serving a majority White student body was associated with greater verbal memory scores (0.09 [0.01, 0.16]). Conversely, each additional year spent in schools serving majority Black or other marginalized students was associated with lower verbal memory scores (−0.16 [−0.31, −0.01]) (Figure 1; Table S3).

Figure 1. Associations between school racial and ethnic composition (in years) and late-life health outcomes.

Data drawn from the Health and Retirement Life History Surveys 2015–2019, N=315. Baseline memory function (0–20) composed of immediate and delayed verbal recall from 1996 to 2016. Baseline depressive symptoms (0–8) were measured using a modified version of the Center for Epidemiological Studies Depression Scale, with higher values indicating more depressive symptoms, based on the participants’ first HRS interview. Poor general health was self-reported through a 5-item Likert scale, response “Poor” and “Fair” were grouped together, while the reference category represented better general health (i.e., Excellent, Very good, Good). Models adjusted for age, gender, birthplace (US South, US Non-South, non-US), birth cohort, and parental education.

School composition and depressive symptoms

Attending schools serving a majority Latinx student body was not associated with depressive symptoms (b [95% CI]= −0.03 [−0.09, 0.04]). Each additional year spent in schools serving a majority White student body was associated with fewer depressive symptoms (−0.05 [−0.11, 0.00]), but estimates were imprecise. Conversely, each additional year spent in schools serving majority Black or other marginalized students was associated with more depressive symptoms (0.18 [0.07, 0.29]) (Figure 1; Table S3).

School composition and poor general health

Each additional year spent in schools serving a majority Latinx student body was associated with a higher risk of poor general health (risk ratio [95% CI]=1.05 [1.01, 1.09]). In contrast, each additional year spent in schools serving a majority White student body was associated with a lower risk of poor general health (0.91 [0.87, 0.96]). Conversely, each additional year spent in schools serving majority Black or other marginalized students was associated with a higher risk of poor general health (1.06 [1.00, 1.11]) (Figure 1; Table S3).

Sensitivity analysis

When estimates were reported separately for time spent in schools serving a majority Black student body or another composition (e.g., a majority composed of a different racial/ethnic group or no single majority group), some associations were higher in magnitude than our primary findings, depending on the specific school composition type. However, these estimates were less precise (e.g., for verbal memory: (b [95% CI]: −0.25 [−0.52, 0.01]) for schools serving majority Black student body) and confidence intervals largely overlapped with the attenuated, more precise primary estimates (Figure S2; Table S3). Associations were similar following additional adjustment for participants’ educational attainment (Figure 2; Table S4) and public-school status (Table S5). Analyses restricted to participants who attended the entirety of their schooling in the US yielded results that were consistent with the primary findings, albeit slightly less precise given the reduced sample size (Table S6). After including selection weights results remained consistent (Table S7).

Figure 2. Associations between school racial and ethnic composition (in years) and late-life health outcomes, additionally adjustment for education.

Data drawn from the Health and Retirement Life History Mail Surveys 2015–2019, N=315. Baseline memory function (0–20) composed of immediate and delayed verbal recall from 1996 to 2016. Baseline depressive symptoms (0–8) were measured using a modified version of the Center for Epidemiological Studies Depression Scale, with higher values indicating more depressive symptoms, based on the participants’ first HRS interview. Poor general health was self-reported through a 5-item Likert scale, response “Poor” and “Fair” were grouped together, while the reference category represented better general health (i.e., Excellent, Very good, Good). Covariate set A included age, gender, birthplace (US South, US Non-South, non-US), birth cohort, and parental education. Covariate set B additionally included educational attainment.

DISCUSSION

In this population-based study of Latinx adults aged 50 and older, we examined the relationship between the racial and ethnic composition of schools and several dementia risk factors, including verbal memory function, depressive symptoms, and general self-rated health. Each additional year spent in a school serving a majority Latinx student body was not significantly associated with verbal memory function or depressive symptoms in later life, but was associated with an increased risk of poorer general health. In contrast, attending schools serving a racially/ethnically discordant student body that was predominantly comprised of a marginalized population (i.e., Black, other, or various racial/ethnic groups) was associated with worse outcomes across all three late-life health indicators. Attending schools serving a majority White student body was associated with better outcomes across all three late-life health indicators. These results were not explained by differences in participants’ educational attainment, suggesting that other mechanisms should be examined in larger samples. This is the first study we are aware of to examine the long-term relationship between school segregation and adult health outcomes among Latinx individuals in the US.

Our findings contribute to the growing body of research examining the short and long-term impacts of segregated schooling on Latinx students.^33–36,63 Given prior reports that majority-minority schools typically receive fewer resources, it is surprising that we did not observe adverse associations between longer durations in schools serving majority Latinx students and verbal memory function, whereas such attendance in majority Black schools was associated with lower scores. This is particularly notable in contrast to our finding that longer attendance at schools serving predominantly non-marginalized populations (e.g., White student majority) was associated with improved verbal memory performance. These patterns suggest that school racial composition may impact cognitive health through mechanisms beyond inequitable resource allocation. One possible mechanism may be bilingualism. Prior research highlights cognitive advantages of bilingualism in older adults, including enhanced working memory.^64,65 Bilingual interactions are likely more prevalent in majority Latinx schools, where students engage with multiple languages across academic and informal settings. These experiences may help buffer the cognitive impacts of inequitable funding and resources, systemic challenges more pronounced in schools serving other marginalized populations (e.g., Black student majority), which were linked to lower memory scores in our study.

Unlike prior research among older Black adults,³⁸ we did not observe long-term psychosocial benefits for Latinx older adults who attended race-concordant schools, and observed harmful associations for self-rated health. Among younger Latinx populations, evidence on the psychosocial impacts of school racial concordance has been mixed, in part due to variation in the timing and measurement of outcomes. Our findings differ from research that has documented short-term psychosocial benefits for Latinx students, such as reduced loneliness and higher self-worth.³⁶ Additionally, we observed fewer depressive symptoms among participants who attended schools serving non-marginalized groups (majority White). However, Dudovitz et al. (2021)³⁴ found in a nationally representative sample of Latinx adolescents that attending a high school with a higher proportion of White students was not significantly associated with depressive symptoms and self-rated health in young or middle adulthood. This discrepancy may reflect differences in the effects of school composition across the lifecourse or cohort differences between the Latinx adults in the HRS who completed the LHMS and younger cohorts. Historical contexts, including xenophobic attitudes towards Latinx individuals in the early 20^th century and the rising tensions and riots in the 1960s-1970s —key moments in the broader civil rights struggle of Latinx communities in the US—likely shaped racialized schooling environments in the HRS-LHMS sample.

In contrast, attending schools serving majority Black students was associated with worse outcomes across all three health measures, likely reflecting the long-standing effects of systemic disinvestment in schools serving predominantly Black students—such as fewer resources, larger class sizes, and underpaid staff.^12,39,43 This contrasts with the health advantages observed among Latinx individuals who attended majority White schools. It could be that individuals attending these schools, which were often better resourced, may have experienced greater access to social capital within higher SES networks, potentially contributing to improved mid-life socioeconomic status, which has been associated with lower depressive symptoms and subsequent health benefits.⁶⁶

These associations may also reflect colorism, with lighter-skinned Latinx individuals more likely to access and benefit from better-resourced, privileged environments due to less discrimination and greater societal acceptance.⁶⁷ Conversely, darker-skinned individuals may face compounded disadvantages, including teacher bias and internalized racism. Future studies should examine both mid-life socioeconomic status, skin tone, and other phenotypic or cultural characteristics as mediators or moderators of the association between school composition and late-life health outcomes among Latinx adults.

This study has several limitations. First, due to our modest sample size, we had limited statistical power. This motivated analytical decisions to examine a cumulative measure and collapse specific school composition types despite inherent construct differences between the individual groupings. For example, only 85 participants in our analytical sample (27%) moved between school composition types, precluding study of more nuanced school composition trajectories. Although our precision was reduced, we still observed that varying school composition types were meaningfully associated with health for older Latinx adults. Second, because the exposure in this study is specific to racialized experiences within the US schooling context, our sample was limited to individuals who had attended school in the US (i.e., Latinx participants born in the US or who migrated by the age of 17). However, results remained consistent when we ran a sensitivity analysis among participants who had their entire schooling experience in the US.

Third, our analyses rely on respondents who survived and completed the LHMS, which may be differential with respect to early-life schooling and/or late-life health. We incorporated weights to account for selection into the LHMS, conditional on being in the core HRS, although other forms of selection bias (e.g., selective survival) could still be a concern. Additionally, school composition was reported retrospectively, raising the possibility of dependent measurement error (Appendix S2).⁶⁸ However, prior studies show high validity in the retrospective assessments collected from the LHMS questionnaire.⁶⁹

Lastly, the overall sample of Latinx participants who participated in the LHMS is not representative of the broader US population of Latinx older adults (e.g., it includes individuals with higher educational attainment, predominantly with Mexican ancestry), limiting the generalizability of our findings. However, to our knowledge, well-established cohorts of older Latinx adults, such as the Hispanic Established Populations for the Epidemiologic Studies of the Elderly and the Hispanic Community Health Study/Study of Latinos-Investigation of Neurocognitive Aging, have not collect data on early schooling experiences (beyond educational attainment) or early residential histories. Latinx individuals are not a monolithic group; as such, this warrants further exploration, particularly at the intersections with other identities, including race, nationality, gender, and class. These limitations underscore the need for more inclusive survey data collection that better reflect the wider Latinx diaspora in aging studies. Given these data gaps, future research should replicate these findings with larger, more representative samples that can allow for evaluating dynamic school-composition trajectories, assessing heterogeneity (e.g., time, region), and examining mid-life mechanisms.

Despite these limitations, our findings are the first, to our knowledge, to evaluate whether racial and ethnic school composition was associated with late-life dementia risk factors among Latinx adults. Specifically, we observed that attending schools serving majority Black students was associated with poorer late-life health while attending schools serving majority White students was associated with better late-life health. Findings were mixed for attending majority Latinx schools, potentially reflecting countervailing influences of attending schools that may have been less well-resourced but may also offer a greater sense of belonging. Given the persistence of de facto racial segregation in public schools,³² policies and programs should prioritize equitable resource distribution across schools and efforts to reduce interpersonal discrimination. Fundamentally, there is a critical need for more data. Researchers should consider collecting information on nuanced educational experiences across all racial and ethnic groups, including those related to segregation, belonging, colorism, resource availability, and beyond. Such data will help us better understand how education shapes dementia risk and dementia inequities in older marginalized groups.

Data Availability Statement:

Data for the Health and Retirement Study is publicly available.