华东师范大学学报(教育科学版)

华东师范大学学报(教育科学版) ›› 2025, Vol. 43 ›› Issue (12): 16-33.doi: 10.16382/j.cnki.1000-5560.2025.12.002

• 高等教育 • 上一篇    下一篇

省级政府高等教育评价改革政策内容再生产及其影响机制研究——基于机器学习与定性比较分析的实证研究

陆根书1,2, 董宇婧1, 刘扬云3   

  1. 1. 西安交通大学人文学院，西安 710049
    2. 西安交通大学中国西部高等教育评估中心，西安 710049
    3. 中国教育发展战略学会教育评价专业委员会，北京 100084
  • 接受日期:2025-08-04 出版日期:2025-12-01 发布日期:2025-11-27
  • 基金资助:
    国家社会科学基金教育学一般项目：“大学毕业生教育-职业错配及其收入效应：多维视角的研究”(BIA240161)。

Policy Reproduction of Provincial Governments’ Higher Education Evaluation Reforms and the Influence Mechanisms: An Empirical Research Based on Machine Learning and Qualitative Comparative Analysis

Genshu Lu1,2, Yujing Dong1, Yangyun Liu3   

  1. 1. School of Humanities and Social Sciences, Xi’an Jiaotong University, Xi’an 710049, China
    2. Western China Higher Education Evaluation Center, Xi’an Jiaotong University, Xi’an 710049, China
    3. Professional Committee on Educational Evaluation, Chinese Society of Educational Development Strategy, Beijing 100084, China
  • Accepted:2025-08-04 Online:2025-12-01 Published:2025-11-27

RichHTML

12

PDF

24

摘要：

高等教育评价改革是增强中国高等教育综合实力和建设高等教育强国的关键环节，省级政府在制定高等教育评价改革政策的过程中对中央政策内容的再生产对当地高等教育评价改革实践将产生重要影响。通过机器学习构建了省级政府政策内容再生产系数，对我国31个省级政府的高等教育评价改革政策内容再生产程度进行了测算，并进一步运用模糊集定性比较分析方法(fsQCA)揭示了省级政府高等教育评价改革政策内容再生产的影响机制。研究发现：省级政府高等教育评价改革政策内容再生产程度总体偏低。省级政府高等教育评价改革政策内容再生产程度高的组态有五种，可进一步概括为三类路径组合，即压力传导型、横向竞争型和自主优化型。研究也发现省级政府政策内容再生产程度不高的组态有三种，可进一步概括为二类路径组合，即资源缺乏型和压力异化型。通过对比政策内容再生产程度高与不高的路径发现，省级政府教育财政投入是省级政府推进高等教育评价改革政策内容再生产的经济基础，省高等教育发展综合实力是省级政府推进高等教育评价改革政策内容再生产的现实基础。省级政府教育财政投入与省级政府的注意力配置存在替代效应，二者均可以对外部压力进行积极回应；不存在外部压力时，其与省高等教育发展综合实力存在协同作用，二者的配合可以有效提升省级政府的政策内容再生产程度。因此，省级政府应激发内生动力，合理配置教育财政投入，因地制宜地贯彻落实中央政策；中央政府则应进一步加强督导检查及奖惩力度，以更好地推动相关政策落实落地。

关键词: 高等教育评价改革政策, 政策内容再生产, 机器学习, fsQCA

Abstract:

Higher education evaluation reform serves as a critical component in enhancing China’s comprehensive higher education capacity and establishing the nation as a global higher education powerhouse. Provincial governments’ policy reproduction during the formulation of evaluation reform policies significantly impacts local practices. Based on the machine learning method, a Doc2vec model was constructed to conduct quantitative calculations on the reproduction of higher education evaluation reform policy content in China’s provinces. Furthermore, Qualitative Comparative Analysis was employed to uncover the influence mechanisms of provincial governments’ policy reproduction in higher education evaluation reform. The research finds that the degree of policy reproduction in provincial higher education evaluation reforms remains generally low. Five high-reproduction configurations were identified, which can be further categorized into three pathways: pressure-driven, horizontal competition-driven, and autonomous optimization. Three low-reproduction configurations were identified, which can be further generalized into two distinct pathway typologies: resource-deficient type and pressure-alienated type. A comparative analysis of pathways with high and low reproduction reveals that the provincial government’s education fiscal input is the economic basis for the provincial government to promote the reproduction of policy content of higher education evaluation reform, and the comprehensive strength of provincial higher education is the realistic basis for the provincial government to promote the reproduction of policy content of higher education evaluation reform. There is a substitution effect between the provincial government’s education fiscal input and the attention allocation of the provincial government, both of which can respond positively to external pressure. When there is no external pressure, they have a synergistic effect with the comprehensive strength of provincial higher education development, and their coordination can effectively enhance the degree of policy content reproduction by provincial governments. Therefore, provincial governments should stimulate internal motivation, rationally allocate education fiscal input, and implement central policies in a way that suits local conditions. The central government should further strengthen supervision, inspection, and reward and punishment measures to better promote the implementation of relevant policies.

Key words: higher education evaluation reform policies, policy reproduction, machine learning, fsQCA