严洁 李钦帅 牛冠捷
摘 要：近年来，有关计算政治学的研究方法、研究逻辑、概念体系和学科边界的讨论一直持续。本文聚焦在计算政治学的研究方法和边界，提出并讨论了政治计算、社会模拟、以及互联网社会实验三种研究方法的分析思路、优势与局限。其中，政治计算在数据量与数据维度方面更具优势；社会模拟对于复杂系统具有更高的适配性；互联网实验是验证因果关系的有力手段，但这些研究方法也会受到数据和模型限制。当前阶段，计算政治学在方法论，研究主题，以及伦理等三个方面也具有明显的边界：作为一种政治学方法论，它依托于政治学理论而存在，不适合于政治哲学问题、战略性问题以及个别事件的研究，社会伦理也为计算政治学者对数理结果的解读和应用划定了边界。
关键词：计算政治学；大数据；政治计算；社会模拟；实验
【作者】严洁，北京大学政府管理学院副教授，博士生导师，北京大学国家治理研究院研究员，北京大学公共治理研究所副所长;李钦帅、牛冠捷，北京大学政府管理学院博士研究生。
【Title】From Quantification to Computation: The Methodology and Boundary of Computational Political Science
Abstract：In recent years, political scientists have been heatedly discussing the research methods, logic, concepts, and application boundaries of computational political science. In this paper, we focus on the research methods and application boundaries of computational political science. We come up with and discuss advantages and disadvantages of three main analytical methods of computational politics: political computation, social simulation, and Internet-based experiment. We argue that political computation has advantages in data volume and dimension; social simulation is more compatible with complex systems; Internet-based experiment is a powerful way to verify causality. However, these analytical methods can be constrained by data and models. The application boundaries of computational politics methods can be delineated by methodology, research topic, and ethics. As a political science methodology, computation political science must be based on solid political theories. Furthermore, computational political science is not a suitable methodology for research questions in political philosophy, grant strategy, and isolated incidents. Moreover, social ethics also circumscribe how political scientists interpret and apply mathematical results.
Key words: Computational political science; Big data; Political computation; Social simulation; Experiment