The effect of attending schools with higher ability peers on memory and language abilities in later‐life

Background Peers influence adolescent behavior, educational, and health outcomes. Evaluating the influence of school‐mates’ cognitive aptitude on later‐life cognition is important for understanding the role of school segregation and tracking in establishing inequality. Yet, few studies have the data to study this. We used a national high‐school based study followed up 58 years later to examine if attending schools with more skilled peers predicts individual later‐life cognition. Method In 1960, about 5% of US high schools participated in Project Talent (n=377,000), where students reported demographics and completed aptitude tests. Some of these students were enrolled in 2018/2019 (mean age=74.6) in the Project Talent Aging Study (PTAS). For each PTAS participant, we calculated peer cognitive aptitude as the average of cognitive aptitudes of all other students who were in the same high school. The outcome, later‐life cognition, was operationalized as z‐score composites for language (semantic and phonemic fluency) and memory (CERAD learning list and delayed recall). For each outcome we estimated multilevel linear models to account for school clustering. All models were adjusted for age, sex/gender, parental education; secondary models were also adjusted for the participant’s own adolescent cognitive aptitude. To examine heterogeneity of the association, we evaluated interactions of peer ability with sex/gender, and the participant’s own adolescent cognitive aptitude. Result Participants were predominantly non‐Hispanic White (80.4%) and women (52.6%). Attending schools with peers who averaged 1 standard deviation higher on cognitive aptitude predicted better later‐life cognition prior to adjustment for own aptitude (language: β=0.162, 95% CI: 0.120, 0.205; memory: β=0.084, 95% CI: 0.041, 0.126), but not after adjustment. However, for women (β=0.045, 95% CI: ‐0.012, 0.101) and those with below mean adolescent cognitive aptitude (β=0.091; 95% CI: 0.016, 0.166), attending schools with higher ability peers predicted better later‐life language scores even after own ability was considered. There was no moderation of these relationships by race/ethnicity. Conclusion People with below average cognitive aptitude in adolescence had better language function 58 years later if they attended high schools with high aptitude peers. Future work will examine the plausible life‐course mechanisms to address why certain groups benefit more from schooling contextual factors.