量化历史研究的过去与未来

On the Past and Future Prospects of Quantitative History Research

近六十年来量化历史研究拓展并加深了我们对历史的认知,使历史研究向科学靠近。不管哪个领域,科学研究的基本流程应该保持一致:首先,提出问题和假说;第二,根据提出的问题和假说去找数据,或者通过设计实验产生数据;第三,做统计分析、检验假说的真伪,包括选择合适的统计分析方法识别因果关系、做因果推断,避免把虚假的相关性看成因果关系;第四,根据分析检验的结果做出解释,如果是证伪了原假说,那原假说为什么错了?如果验证了当初的假说,又是为什么?这里挖掘清楚"因"导致"果"的实际传导机制甚为重要;第五步就是写报告文章。传统历史研究在第二至第四步上做的不够完整。所以,量化方法不是要取代传统历史研究,而是对后者的补充。本文通过一些现有成果说明,量化史学不只是"用数据说话",而是通过统计分析,既可令人信服地证明或证伪现有假说,也可以从历史现象中发现全新的认知。

For almost six decades,quantitative history research has substantially deepened our understanding of history and made historical research more of a science. Regardless the field of inquiry,the basic scientific method follows five steps. First, identify the research question and develop a testable hypothesis; Second,find or generate data through experiments or from pre-existing sources relevant to the hypothesis; Third,conduct statistical analysis and hypothesis testing,including establishing or rejecting causal relations; Fourth,interpret the test results and uncover the channels through which the causality chain runs; Finally,write a research report or paper summarizing the findings and analysis. This paper argues that there is considerable room for improvement in traditional historical research,in particular from the second to the fourth step outlined herein. That is,historians should emphasize the use of both large samples and robust statistical methods,rather than just relying on case studies rather than replacing it;Historians should also make use of quantitative methods to establish causality and identify channels in which the causality runs. This paper shows that quantitative methods can not only make history research more robust and efficient but also discover new knowledge about history that would not be possible using traditional approaches.