Multivariate Interdependencies of Chinese, Mathematics, and English Performance among High-School Students

Abstract

This study aimed to examine interdependencies among Chinese, Mathematics and English performance and to test for a unifying proficiency factor. It analysed longitudinal data from 27,750 students over 22 examination sessions in six years, applying Pearson correlations, multivariate general linear models, cross-lagged panel analysis, principal component analysis and confirmatory factor analysis. Results indicated moderate but significant bivariate associations and reciprocal predictive effects, most notably a leading influence of Chinese on subsequent English performance. Factor analyses consistently identified a single latent construct academic proficiency accounting for 62.9 % of score variance. This study is the first to demonstrate a general academic proficiency factor across all three Gaokao subjects using cross-lagged and factor analytic methods on large-scale longitudinal data. These findings indicate that student performance across Chinese, Mathematics, and English is moderately interrelated and can be parsimoniously represented by a single latent academic proficiency factor. While the study does not directly assess underlying cognitive or motivational traits, the consistent factor loadings and strong model fit suggest the presence of a general academic dimension influencing all three subjects. This suggests that instructional designs which foster cross-subject skill transfer, such as embedding reading strategies in Mathematics or reinforcing logical reasoning in language tasks, may enhance overall academic performance. Furthermore, educational interventions aimed at strengthening general learning processes, such as metacognitive regulation and effort management, could offer broad-based support across disciplines. These insights can inform the development of more integrated and efficient curricular strategies under China’s New Gaokao framework and may have relevance for other high-stakes assessment systems.