大数据时代的因果推断——教育政策评估的新路径  

Causal Inference in the Age of Big Data:A New Approach to Education Policy Evaluation

作　　者：郭娇 吴寒天 GUO Jiao;WU Hantian(Institute of Higher Education,East China Normal University,Shanghai 200062,China;College of Education,Zhejiang University,Hangzhou 310058,China)

机构地区：[1]华东师范大学高等教育研究所,上海200062 [2]浙江大学教育学院,杭州310058

出　　处：《重庆高教研究》2022年第4期39-48,共10页Chongqing Higher Education Research

基　　金：教育部人文社会科学研究规划基金项目“基于Trace-SRL的本科生在线学习行为分析”(21YJA880014)。

摘　　要：在研究方法与数据来源不断更新迭代的当下,对新兴路径与范式的探索业已成为教育政策研究者亟须开展的工作。蓬勃发展的“数据密集型科学发现”被界定为科学方法革命的“第四范式”,表征科学探究的基本范式在当前大数据时代前所未有的变化。这一深刻变革不仅发生在自然科学与工程领域,也同样作用于以教育学为代表的广义社会科学领域。然而,这一新范式与科学研究中既有的因果推断逻辑似乎存在一定程度的互斥,如何将二者有机融合或可被视为中观及宏观维度教育政策研究亟待探索的“蓝海”。聚焦教育政策评估中既有的因果推断范式,系统梳理国外运用随机实验、准实验设计以及机器学习等方法在该领域所开展的前沿研究,继而探讨如何构建“大数据-因果推断”这一新型研究路径,为教育政策实施效果评估提供依据。With the continuous updating and iteration of research methods and data sources,the exploration of emerging paths and paradigms has become an urgent work for the educational policy researchers.The booming“data-intensive scientific discovery”is defined as the“fourth paradigm”of the scientific method revolution,which represents the unprecedented change of the basic paradigm of scientific inquiry in the current“big data era”.The profound change takes place not only in the field of natural science and engineering,but also in the field of broad social sciences represented by pedagogy.However,there seems to be a certain degree of mutual exclusion between the new paradigm and the existing causal inference logic in scientific research.How to organically integrate and apply the two can be regarded as the“blue ocean”to be explored in the study of meso and macro educational policies.Focusing on the existing causal inference paradigm in education policy evaluation,the frontier research carried out in the field was systematically clarified by using random experiment,quasi experimental design and machine learning,and then how to build a new research path of“big data causal inference”was discussed,so as to provide a basis for the evaluation of the implementation effect of education policy.

关 键 词：大数据 因果推断 教育政策 政策评估 科学研究范式 

分 类 号：G640[文化科学—高等教育学]