Benefits of educational attainment on adult fluid cognition: international evidence from three birth cohorts

Abstract

Background: Educational attainment is highly correlated with social inequalities in adult cognitive health; however, the nature of this correlation is in dispute. Recently, researchers have argued that educational inequalities are an artefact of selection by individual differences in prior cognitive ability, which both drives educational attainment and tracks across the rest of the life course. Although few would deny that educational attainment is at least partly determined by prior cognitive ability, a complementary, yet controversial, view is that education has a direct causal and lasting benefit on cognitive development.

Methods: We use observational data from three birth cohorts, with cognition measured in adolescence and adulthood. Ordinary least squares regression was used to model the relationship between adolescent cognition and adult fluid cognition and to test the sensitivity of our analyses to sample selection, projection and backdoor biases using propensity score matching.

Results: We find that having a university education is correlated with higher fluid cognition in adulthood, after adjustment for adolescent cognition. We do not find that adolescent cognition, gender or parental social class consistently modify this effect; however, women benefited more in the 1946 sample from Great Britain.

Conclusions: In all three birth cohorts, substantial educational benefit remained after adjustment for adolescent cognition and parental social class, offsetting an effect equivalent of 0.5 to 1.5 standard deviations lower adolescent cognition. We also find that the likelihood of earning a university degree depends in part on adolescent cognition, gender and parental social class. We conclude that inequalities in adult cognition derive in part from educational experiences after adolescence.

Figure 1

Graphical representation showing cognitive selection into educational attainment and educational benefits on adult cognition. Secondary Qualifications (SQ) lines are solid; University Degree (UD) lines are dashed. Dark grey bands represent confidence intervals. Light grey bands show regions where different qualifications have graduates with similar adolescent intelligence. (A) Cognitive processes: adolescent and adult cognition are correlated; adolescent intelligence defines propensity for educational attainment with no educational benefits. (B) Educational benefits: no cognitive selection for educational attainment; adult cognition determined by both adolescent intelligence and educational benefits. (C) Multiple processes: evidence for both cognitive selection into education and benefit of education on adult cognition

Figure 2

Adult fluid cognition by adolescent intelligence for those who were educated to a university degree (UD) as compared with those with secondary qualifications (SQ). We provide OLS estimates using thick black lines, and 95% confidence intervals using mirrored solid thin grey lines. For explanatory purposes, we use horizontal solid lines capped by circles and dashed vertical lines to estimate a point estimate for the cognitive offset evaluated at the sample average (ΔC₀). (A) In the U.S. 1939 cohort, ΔC₀ is large at 1.49 (1.44–1.54; P < 0.001). (B) In the 1946 G.B. cohort, ΔC₀ was 0.40 (−0.01 to 0.81; P > 0.05). (C) In the 1958 G.B. cohort, ΔC₀ was 0.49 (0.45–0.53; P < 0.001)

Figure 3

Adolescent cognition related to adult fluid cognition for women and men. Results for women are grey, whereas those for men are black. In both G.B. cohorts, women had higher adult fluid cognition than did men. (A) In the U.S. 1939 cohort, no gender differences were evident. (B) In the G.B. 1946 cohort, women benefited twice as much (P < 0.05) from higher education [ΔC₀ = 1.05 (0.86–1.24)] as men [ΔC₀ = 0.46 (0.28–0.64)]. (C) In the 1958 G.B. cohort, women and men benefited equally