Mass higher education has become a global trend in educational reform (Alves & Tomlinson, 2021), bringing with it issues such as resource differentiation and equity (Carpentier, 2021). In Europe, countries like Germany and the UK have implemented government-led resource balancing mechanisms to promote fairness and quality in higher education. In contrast, the United States relies on market-driven mechanisms, which result in a high concentration of resources and significant inter-institutional disparities. Following the implementation of the Double First-Class policy, which aims to build world-class universities and disciplines, China’s higher education system has gradually shifted toward development with an emphasis on quality. Under this policy, the crucial issue that needs to be explored is how to balance the expansion of scale with the quality, efficiency, equity, and other related concerns of education.

In this context, this study focuses on universities directly under the Ministry of Education and uses Latent Profile Analysis (LPA) based on key indicators such as student numbers, student-teacher ratio, and per-student funding to reveal the heterogeneity of resource allocation and its evolutionary patterns in these institutions under the Double First-Class policy.

The core question of this study is: What are the characteristics or types of resource allocation in universities under the Ministry of Education within the context of the Double First-Class policy? Are these types evolving over time, or are they becoming entrenched? Furthermore, how can theoretical frameworks be used to analyze the path dependence and policy effects of resource allocation across different universities, thus providing a theoretical basis for policy improvement?

The primary goal of this study is to identify and analyze the resource allocation patterns of universities directly under the Ministry of Education through LPA and to explore the dynamic evolution of resource allocation under policy intervention. The research aims to address the following points: 1. Identify the heterogeneity of resource allocation in terms of per-student funding, student scale, and student-teacher ratio in universities before and after the policy implementation; 2. Investigate the impact of the Double First-Class policy on resource allocation and analyze the changes in resource allocation and their driving factors post-implementation; 3. Reveal the path dependence characteristics formed through long-term resource allocation and explore how policy either promotes or constrains the formation of this path dependence in different universities.

This study constructs a conceptual framework based on path dependence theory. Path dependence theory emphasizes that historically established resource allocation models can profoundly influence the effectiveness of current and future policies (David, 2007). In the field of higher education, path dependence in resource allocation could be evident. Once a particular resource allocation model is established, it tends to be further reinforced by factors such as inter-university competition and institutional constraints, making it difficult to change. Path dependence not only reflects the inertia of resource allocation in universities but also reveals potential institutional barriers that may arise during educational policy reforms. This study will analyze the evolution of resource allocation characteristics of universities in combination with path dependence theory.

This study uses a sample of 75 universities directly under the Ministry of Education of China, with data drawn from each institution’s publicly available teaching quality reports, financial statements, and statistical data from the Ministry of Education for the years 2015 and 2023. The study focuses on the dynamic evolution of educational resource input and allocation before and after the implementation of the Double First-Class policy. The selected universities are directly managed by the Ministry of Education, making them most susceptible to policy effects. The study selects three core indicators—student numbers (measuring institutional scale), student-teacher ratio (reflecting the efficiency of faculty allocation), and per-student funding (representing the intensity of resource input)—to construct a closed-loop analytical framework that covers “scale-process-input”. In terms of methodology, Latent Profile Analysis is employed to identify heterogeneous patterns of resource allocation in these universities. LPA is a person-centered analytical technique (Pastor et al., 2007; Hickendorff et al., 2018) that has been widely applied in psychology and personality research (Magson et al., 2022; Specht et al., 2014; Jansen In De Wal et al., 2014) and occasionally used in funding studies (Zang et al., 2020). In this study, LPA analysis is conducted using RStudio, with model fit assessed by examining the goodness of fit for latent profiles ranging from 1 to 6 classes. The optimal number of profiles is determined using criteria such as the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Entropy.

The study identified three types of universities in both 2015 and 2023: “high-resource-intensive,” “balanced development,” and “scale expansion-inefficient.” The first type has high per-student funding, low student-teacher ratio, and a medium student scale. The second type features moderate funding and student-teacher ratio, with smaller student numbers. The third type has low funding, high student-teacher ratio, and large student numbers. In 2015, the proportions of these types were 4.00%, 21.33%, and 74.67%, respectively, and in 2023, they were 5.33%, 25.33%, and 69.33%, showing notable stability. The Double First-Class policy led to a slight increase in high-resource universities (+1.33%), but their per-student funding decreased by 5.6% due to scale expansion, revealing a conflict between policy incentives and resource allocation efficiency. Balanced development universities, despite a 9.5% reduction in per-student funding, maintained stable student-teacher ratios by limiting their scale growth (only 0.8%), thus becoming a resilient part of the higher education system. Meanwhile, scale expansion-inefficient universities showed worsening path dependence, with their student-teacher ratio rising from 16.11 to 16.23, and per-student funding remaining stagnant at around 70,000 yuan. Although the proportion of these universities dropped by 5.34 percentage points, their dominant position persisted, indicating a “low equilibrium trap” in the resource acquisition model relying on scale expansion. The results highlight path dependence and institutional barriers in resource allocation. Despite some changes in resource allocation patterns, the overall stability reflects the persistence of path dependence. Future policies should emphasize dynamic adjustment and categorized management, promoting more efficient and equitable resource distribution, and avoiding the solidification of inefficient allocation patterns.

Alves, M. G., & Tomlinson, M. (2021). The changing value of higher education in England and Portugal: Massification, marketization and public good. European Educational Research Journal, 20(2), 176–192. https://doi.org/10.1177/1474904120967574 Carpentier, V. (2021). Three stories of institutional differentiation: Resource, mission and social inequalities in higher education. Policy Reviews in Higher Education, 5(2), 197–241. https://doi.org/10.1080/23322969.2021.1896376 David, P. A. (2007). Path dependence: A foundational concept for historical social science. Cliometrica, 1(2), 91–114. https://doi.org/10.1007/s11698-006-0005-x Hickendorff, M., Edelsbrunner, P. A., McMullen, J., Schneider, M., & Trezise, K. (2018). Informative tools for characterizing individual differences in learning: Latent class, latent profile, and latent transition analysis. Learning and Individual Differences, 66, 4–15. https://doi.org/10.1016/j.lindif.2017.11.001 Jansen In De Wal, J., Den Brok, P. J., Hooijer, J. G., Martens, R. L., & Van Den Beemt, A. (2014). Teachers’ engagement in professional learning: Exploring motivational profiles. Learning and Individual Differences, 36, 27–36. https://doi.org/10.1016/j.lindif.2014.08.001 Magson, N. R., van Zalk, N., Mörtberg, E., Chard, I., Tillfors, M., & Rapee, R. M. (2022). Latent stability and change in subgroups of social anxiety and depressive symptoms in adolescence: A latent profile and transitional analysis. Journal of Anxiety Disorders, 87, 102537. https://doi.org/10.1016/j.janxdis.2022.102537 Pastor, D. A., Barron, K. E., Miller, B. J., & Davis, S. L. (2007). A latent profile analysis of college students’ achievement goal orientation. Contemporary Educational Psychology, 32(1), 8–47. https://doi.org/10.1016/j.cedpsych.2006.10.003 Specht, J., Luhmann, M., & Geiser, C. (2014). On the consistency of personality types across adulthood: Latent profile analyses in two large-scale panel studies. Journal of Personality and Social Psychology, 107(3), 540–556. https://doi.org/10.1037/a0036863 Zang, S., Zhao, M., OuYang, J., & Wang, X. (2020). Interpretation of China’s 2017 health expenditure: A latent profile analysis of panel data. BMJ Open, 10(6), e035512. https://doi.org/10.1136/bmjopen-2019-035512