教育政策定量评估方法中的因果推断模型以及混合方法的启示

Causal Inference Models in Education Policy Evaluation and Mixed Methods

本文从概述的角度介绍了国际上主流的评估教育政策的定量方法——因果推断模型。因果推断模型是借鉴自然科学中实验的理念而演化出来的一整套定量研究方法，包括随机试验、自然实验、断点回归模型、倾向分数配对法等。这些方法主要是为了解决教育领域中的自选择偏误。基于对因果推断模型的介绍，本文总结了定量研究方法的复杂性和局限性，并探讨了混合方法在教育政策评估中的潜在优势。

This article introduces causal inference models in education policy evaluation that are currently used worldwide. Causal inference models are generated from the idea of experiment in natural sciences. These models, including random experiment, natural experiment, regression discontinuity,and propensity score matching, aim to address the problem of self-selection bias. Based on the discussion of causal inference models, this article summarizes the complexity and limitations of quantitative methodology, and discusses the potential strength of mixed methods in education policy evaluation.