Regression Discontinuity Design: Theoretical Developments and Extended Applications

As one of the most important quasi-experimental methods, the Regression Discontinuity Design (RDD) has been widely used in empirical researches in recent years, and it has been undergoing a series of rapid theoretical developments and breakthroughs. This paper provides systematic introduction and guidance of two analytical frameworks of RDD, the Continuity-Based Framework and the Local Randomization Framework, including their basic assumptions, estimation methods, statistical inference, implementations and main steps of empirical analyses. This paper then compares the differences of these two analytical frameworks. Furthermore, this paper expounds RDD methods in three special circumstances, including RDD with multiple running variables, RDD with multiple cutoffs, and Regression Kink Design, and offers detailed approaches to their code implementations. These theoretical developments and their implementations can extend the application of RDD to a wider range of fields and increase the feasibility of empirical researches using RDD.