政治科学视角下的大数据方法与因果推论

Big Data Approach and Causal Inference in Political Science

伴随着科技进步和治理实践的不断演进,日新月异的组织制度环境、学科间融合和新兴方法的广泛应用,持续驱动着政治科学理论和方法创新。大数据既是新兴研究方法,为政治学研究提供丰富数据源,更是政治学研究新议题,日益为政治学的知识生产提供新动力。针对政治科学对因果性知识的严格要求,大数据方法被认为重视相关性分析而非因果性研究。近年来,伴随着大数据方法在方法论层面日趋成熟、方法技术日益丰富、研究议题逐步拓展,且与传统社会科学方法逐步融合,大数据方法推进因果推论的能力逐步完善,革新着数据采集、概念测量、相关性分析、因果性与预测性分析等因果推论的各环节,形成了大数据与统计方法、大数据与小数据分析、大数据与实验研究、大数据模拟方法等多种生产和检验因果性知识的方法路径。

With the continuous evolution of technological advancement and governance practice,the rapid shift of organizational and institutional environment,interdisciplinary integration and various applications of new research methods continue to drive the theoretical and methodological innovation of political science.Big data is not only a new methodology,but also provides political scientists rich sources of data.With the rapid expansion of big data related technique in real-world politics,it creates sets of new research topics for understanding political system,and provides the new impetus to the knowledge accumulation of political science.In view of the priority of causality related knowledge in political science,big data approach is regarded as good at exploring correlation rather than examining causality.Recently,based on the gradual maturity of big data methodology,the diversity of data analytics,the widespread coverage of research topics,there is an obvious trend of the deeply integration between big data approach and the traditional social science methods.Therefore,big data approach strengthens its capacity to make causal inference through innovating the methodology of data collection,concept measurement,correlation analysis,causality and prediction.To be summed,there are four emerging paths such as intergrading big data and econometrics,combining big data and small data analysis,big data assisted experimental study,and big data assisted simulation method.