摘要
与建立相关关系的统计模型不同,反事实框架下的因果推论有三大特征:第一,不在模型上强加假定,给方法应用者留下更多解释、诠释和补充的空间。第二,系统性,因果推论的各种具体工具之间存在步步为营、层层后退的关系,需要研究者慎重考虑方法的内外效度与研究议题之间的关系,并在文中以文字形式充分佐证。第三,因果推论不只可以做效应评估,也可以做机制分析,并且特别适用于认知行为的研究。
Unlike correlation-based regression model,the counterfactual framework has three features.First,it leaves more room for solution,interpretation,and supplementation for method users rather than impose assumptions on the model.Second,the step-by-step relationship between the various concrete tools for causal inference decides that researchers should carefully consider the relationship between the internal and external validity of the method.Third,causal inference can not only be used for effect evaluation,but also for mechanism analysis,in particular,the studies on cognitive behavior.
出处
《东南大学学报（哲学社会科学版）》
CSSCI
2020年第4期98-109,157,共13页
Journal of Southeast University(Philosophy and Social Science)
