论自然实验的定义

On Definitions of Natural Experiments

自然实验是社会科学中近年来广为流行的研究方法。在实证研究文献中,研究者较少讨论该方法的定义,而是采用工具性的使用态度。目前存在三种主流的自然实验定义,分别是:利用自然或社会事件、比较法和随机化分组。结合社会科学和医学研究案例,可以说明三种定义各自的适用范围以及局限性。在此基础上本文进一步分析了三种定义之间的关联,并指出定义二更加符合自然实验的适用范围和优势。根据定义二,自然实验不能还原为实验室实验或者观察,而是占据了此二者之间独立的方法论位置。

Recently, natural experiments are getting popular among social sciences. In empirical research literature, researchers take little effort to explain its methodological definition. At present, there are three major approaches defining natural experiments: the usage of natural or social events, comparative method, and randomization. This article demonstrates the scope of application and limitations of each definition with cases in social science and medical studies. After further analysis on the pros and cons of these definitions, this article supports the second definition of comparative method, which is more in line with its actual usage and advantages. According to the second definition, natural experiments can be reduced to neither laboratory experiments nor observation, but deserve an independent methodological role in between.