Associations between school enjoyment at age 6 and later educational achievement: evidence from a UK cohort study

The study investigates whether early school enjoyment predicts later academic achievement, whether family background factors explain this relationship, and whether sex differences in enjoyment contribute to sex differences in achievement. Motivated by the intuition and policy interest that enjoyment of learning is more modifiable than socioeconomic status or cognitive ability, and the paucity of longitudinal evidence from early childhood, the authors examine school enjoyment at age 6 in relation to GCSE outcomes at age 16. The work also assesses socioeconomic and demographic patterning of enjoyment to understand potential confounding, and evaluates whether enjoyment could serve as an early indicator and intervention target to improve educational outcomes.

Prior research generally finds positive links between enjoyment and academic outcomes, though measures vary. Small studies reported that enjoyment of a project or exams predicted higher grades (e.g., Chicago high school study, n=90, enjoyment accounted for almost half a grade difference; Austrian university psychology students, n=388, exam enjoyment associated with a one-sixth grade increase). A meta-analysis showed a negative correlation between boredom and academic outcomes (r ≈ -0.24). Enjoyment of specific subjects (maths, reading) associates with higher subject performance. In large UK samples, school enjoyment at age 14 (LSYPE, ~11,000) and at age 11 predicted later achievement at age 16 and 14 respectively, with stronger effects in mathematics. However, some primary-school-age studies found no strong contemporaneous links. The literature suggests mechanisms via increased motivation and positive emotions that support learning; school-based interventions can increase positive emotions including enjoyment. Measurement inconsistency and combining enjoyment with broader wellbeing constructs often obscure enjoyment’s unique role. Confounding by family socioeconomic position, parental involvement, and other contextual factors is a concern, and many studies lacked adequate controls. Evidence on sex differences is mixed: girls generally report higher school enjoyment and steeper declines into secondary school; boys may enjoy maths more. Limited evidence suggests enjoyment may mediate sex differences in achievement. Overall, there is a lack of longitudinal evidence from early childhood to later standardized outcomes, and uncertainty about socioeconomic confounding.

Design and data: Observational cohort analysis using the Avon Longitudinal Study of Parents and Children (ALSPAC), UK. Pregnancies with expected delivery 04/01/1991–12/31/1992 were enrolled; core sample 14,899 children alive at age 1. Educational outcomes were obtained via linkage to England’s National Pupil Database (NPD). Ethical approvals obtained; informed consent provided. Exposure: School enjoyment measured twice at age ~6: (1) at 6.2 years: “Do you like school?” (yes/no); (2) at 6.5 years: “How much do you like going to school?” (I like it a lot/bit vs I don’t like it). Responses combined into a 3-category variable: no enjoyment at both times; mixed enjoyment (enjoyed at one time); enjoyed at both times. Outcomes: (1) Fine graded, capped GCSE points score at age 16 (end of compulsory schooling); (2) Binary indicator of achieving 5+ A*–C GCSEs including English and Maths. Covariates: Sex; month of birth; ethnicity (white/non-white); cohort school year; age when enjoyment was reported; parental socioeconomic position (highest parental occupational social class, categorized with top and bottom two combined due to low counts); maternal highest education (Degree, A-level, O-level, CSE/vocational); cognitive ability at age 8 via WISC short form; child’s opinion of teacher at age 6 (mother-reported, 4-point scale); temperament at age 6 (composite of seven self-reports); confidence in work quality at age 8 (four items); confidence in intelligence at age 8 (three items); quality of friendship group at age 8 (five items); home learning environment at age 3 (frequency of early learning activities across six items). Missing data handling: Multiple Imputation by Chained Equations (MICE) for all variables except sex, month of birth, and ethnicity; 100 imputed datasets; final imputed sample n=12,135; complete-case n=2,544. Auxiliary variables were included to support the Missing At Random assumption and reduce selection bias. Statistical analysis: (a) Multinomial logistic regression assessed associations of socioeconomic/demographic and child factors with school enjoyment categories. (b) Linear regression estimated associations between school enjoyment and GCSE points. (c) Logistic regression estimated odds of achieving 5+ A*–C including English and Maths. Three model specifications were applied: Model 1 adjusted for sex, month of birth, ethnicity, cohort year, and age at enjoyment reporting; Model 2 additionally adjusted for parental occupational class and maternal education; Model 3 additionally adjusted for cognitive ability, home learning environment, opinion of teacher, temperament, confidence in work, confidence in intelligence, and friendship quality. Primary inferences used multiply imputed results; complete-case analyses provided as sensitivity checks.

Descriptives and patterning: - In the imputed sample (n=12,135), 77.7% reported enjoying school at both time points, 16.0% had mixed enjoyment, and 6.28% did not enjoy school at either time. - No clear socioeconomic patterning of school enjoyment by parental occupational class or maternal education; degree-educated mothers had a modestly higher (borderline) association with enjoyment at both times (OR 1.36; 95% CI 0.99, 1.87). - Higher cognitive ability associated with higher odds of enjoying school (fully adjusted OR 1.16; 95% CI 1.02, 1.32 per SD). - Girls were about twice as likely to report enjoying school as boys (adjusted OR 1.97; 95% CI 1.56, 2.48). Non-white children were almost twice as likely to report enjoyment as white children (OR 1.87; 95% CI 0.99, 3.55). Being born later in the school year associated with lower enjoyment in unadjusted models (OR 0.95; 95% CI 0.92, 0.99), attenuating after covariate adjustment (OR 0.98; 95% CI 0.94, 1.01). - Strong associations with proximal child factors: very positive opinion of teacher (OR 9.44; 95% CI 4.45, 20.02), temperament (OR 1.19; 95% CI 1.12, 1.26), and work confidence (OR 1.36; 95% CI 1.27, 1.45). Achievement outcomes (GCSE points): - Relative to no enjoyment, mixed enjoyment associated with β 12.53 (95% CI 2.42, 22.65) unadjusted; attenuated to β 8.05 (0.22, 15.89) with socioeconomic adjustment; and β 6.16 (−1.70, 14.02) in the fully adjusted model. - Enjoyed school associated with β 29.34 (19.40, 39.29) unadjusted; β 17.75 (10.45, 25.05) with socioeconomic adjustment; and β 14.41 (6.90, 21.93) fully adjusted. Approximate magnitudes equate to about five grades total (unadjusted) and two grades (fully adjusted) across best eight GCSEs (6 points per grade). - In fully adjusted models, effect size of enjoyment was similar to sex (β 21.61; 95% CI 18.73, 24.48) and the largest social class contrast (β 22.19; 95% CI 17.49, 26.89). Maternal degree vs CSE/vocational: β 43.84 (38.37, 49.31). Cognitive ability per SD: β 43.01 (41.25, 44.77). Achievement outcomes (5+ A*–C including English and Maths): - Mixed enjoyment OR 1.28 (95% CI 1.02, 1.60) unadjusted; OR 1.27 (0.97, 1.66) socioeconomic-adjusted; OR 1.18 (0.90, 1.55) fully adjusted. - Enjoyed school OR 1.64 (1.33, 2.03) unadjusted; OR 1.48 (1.16, 1.90) socioeconomic-adjusted; OR 1.29 (0.99, 1.67) fully adjusted. - Cognitive ability was the strongest predictor: per SD increase OR ≈ 3.10 (95% CI 2.88, 3.34). Other covariates: - Home learning environment at age 3 associated with higher achievement (β 3.34; 95% CI 2.06, 4.61; OR 1.07; 95% CI 1.03, 1.12). - Confidence in intelligence at age 8 associated with higher achievement (β 3.62; 95% CI 2.65, 4.60; OR 1.09; 95% CI 1.05, 1.13). - Limited evidence that teacher opinion, temperament, confidence in work, or friendship quality (measured at 6–8 years) were independently associated with achievement after adjustment. Interactions and robustness: - Little evidence for interactions between enjoyment and social class, maternal education, cognitive ability, or sex on achievement (complete-case interaction p-values > 0.08), with limited power noted. Results were broadly consistent between imputed and complete-case analyses; complete-case tended to overestimate enjoyment–achievement and underestimate social background and cognitive ability associations.

Early school enjoyment at age 6 is positively associated with educational achievement at age 16, even after accounting for socioeconomic background, cognitive ability, and multiple child-level factors. The persistence of associations after extensive adjustment and the lack of strong socioeconomic patterning suggest that enjoyment captures a distinct, potentially modifiable aspect of the learning experience, beyond innate ability or family background. Given that enjoyment differences are almost as large as those for sex and social class, enjoyment may serve as an early indicator for pupils who could benefit from additional educational support. The findings also point to the potential of early interventions that enhance positive emotions and engagement in schooling, as enjoyment tends to decline with age. While sex differences in enjoyment are marked (girls higher), there was limited evidence that the enjoyment–achievement association differs by sex in this sample; larger studies are needed. Relative age effects were observed for achievement but not clearly for enjoyment, suggesting that later-born children’s lower achievement does not stem from lower enjoyment in this cohort. Overall, results reinforce the importance of affective factors in early education and encourage further research into causal mechanisms and effective strategies to foster enjoyment.

School enjoyment at age 6 strongly associates with GCSE achievement at age 16, with effect sizes comparable to key sociodemographic inequalities and robust to extensive covariate adjustment. Enjoyment appears only weakly patterned by socioeconomic background, indicating it may reflect modifiable aspects of early educational experiences. The study highlights school enjoyment as a promising focus for early identification and intervention to improve long-term educational outcomes. Future research should replicate these findings in representative and diverse settings, elucidate mechanisms (including reasons for low enjoyment), assess school/class-level influences, and test interventions—ideally experimentally—to determine whether enhancing early enjoyment can causally improve attainment and reduce inequalities.

- Generalisability: ALSPAC is not fully representative of the UK school population (underrepresentation of single-parent families, renters, some ethnic minorities), limiting external validity. - Missing data and imputation: Results rely on Multiple Imputation by Chained Equations with assumptions of Missing At Random; bias may occur if important predictors of missingness were omitted, despite inclusion of auxiliary variables. Complete-case analyses suggested potential selection bias and different magnitudes. - Observational design: Potential residual confounding cannot be excluded; associations may not be causal. - Temporal ordering and collider bias: Some covariates (cognitive ability at 8, confidence in work/intelligence, friendships) were measured post-exposure and may introduce collider bias; cognitive ability is relatively stable, but bias cannot be ruled out. - School/class effects: Insufficient clustering and nonrepresentative school sampling precluded multilevel analysis of school- or class-level determinants of enjoyment. - Measurement limitations: Child’s liking of teacher was mother-reported and may be mismeasured; enjoyment measures were coarse (binary per timepoint), possibly attenuating true associations. - Reasons for low enjoyment: Data did not allow investigation of underlying causes (e.g., health, social-emotional/behavioral difficulties), limiting implications for targeted interventions.