Meta-analysis of cognitive, emotional, and behavioral predictors of academic achievement in higher education students

Higher education outcomes are frequently associated with cognitive ability and the outcomes of previous school years, but there are likely emotional and behavioral factors that contribute to these as well. This meta-analysis examined the relative effects of cognitive, emotional, and behavioral predictors on academic performance in university and college students.

Seventeen studies with higher education students were included. Eligible studies reported on at least one cognitive predictor (e.g., prior GPA, working memory, metacognition), emotional predictor (e.g., emotional intelligence, achievement emotions, emotion regulation), or behavioral/self-regulatory predictor (e.g., study habits, time management, engagement, procrastination) of one of the outcomes (GPA, course grades, or pass/fail status). Random effects models were used to transform effect sizes to Fisher's z and pooled. Subgroup and meta regression analyses were used to investigate heterogeneity and moderators (design, field, region, measurement characteristics) while multiple diagnostics were used to assess publication bias.

The cognitive domain showed the strongest pooled association with academic achievement, r = 0.197, 95% CI [0.037, 0.347]. Emotional predictors showed a weak and imprecise association with academic achievement, r = -0.105, 95% CI [-0.376, 0.182]. Overall, the pooled effect across all predictors was small and not statistically strong, r = 0.017, 95% CI [-0.170, 0.202], with very high heterogeneity. Behavioral predictors showed an unstable estimate, r = 0.002, 95% CI [-0.477, 0.481], and should be interpreted cautiously.

Taken together, the corrected synthesis suggests that cognitive predictors are the most consistent domain, whereas emotional and behavioral findings are less stable and require cautious interpretation. Cognitive predictors appear to be the most reliable domain in the current evidence base, while emotional and behavioral predictors remain tentative. Therefore, interventions that integrate cognitive assistance with specific training in SRL and emotion management are more in line with the evidence than interventions based on performance only.