Journal of Higher Education Research

Recent educational assessment reforms in China mandate a fundamental shift from a singular focus on test scores toward comprehensive, diagnostic evaluations that support student development. However, a significant gap persists between these policy ambitions and current practices, which remain predominantly reliant on variable-centered analyses of composite scores. This article proposes a paradigm shift from a variable-centered to a person-centered analytical framework. We advocate for the application of two advanced statistical techniques, Latent Profile Analysis (LPA) and Multilevel Latent Class Analysis (MLCA). LPA allows for the identification of distinct, meaningful subgroups of students based on their multidimensional profiles of abilities and knowledge structures, moving beyond mere ranking to reveal heterogeneous learning patterns. Building on this, MLCA extends the analysis to account for the nested structure of educational data, enabling the simultaneous identification of latent typologies at both the individual and group levels. This integrated approach facilitates a more nuanced understanding of academic proficiency, providing a robust methodological foundation for generating diagnostic “student portraits” and “school profiles”. The article concludes that this person-centered, multilevel framework is not only a technical advancement but also a strategic imperative for aligning educational assessment with the national reform agenda, ultimately fostering equitable and high-quality education.