Regression Discontinuity Designs

Instructor(s):

  • Rocio Titiunik, University of Michigan
  • Sebastian Calonico, University of Miami

This three-day workshop gives an introduction to the basic principles of the Regression Discontinuity (RD) design, and also discusses recent methodological developments in the interpretation and analysis of this quasi-experimental design. The workshop provides an accessible summary of the assumptions behind the RD design and introduces participants to different methods that are appropriate for the successful analysis of RD empirical applications.

The course introduces the parameters of interest in RD designs and focuses on different methods to perform statistical inference, including flexible global parametric inference, local-polynomial non-parametric inference, and local randomization-based inference. When discussing local non-parametric inference, special attention is given to bandwidth selection methods and calculation and computation of valid confidence intervals. It also discusses how to design and implement falsification tests. Additionally, several extensions to the basic RD design are discussed, including multi-cutoff and geographic RD designs, Regression Kink designs, and RD designs with discrete running variables.

Every day, part of the class will be devoted to solve hands-on computer-based exercises, all of which will be based on previously published "real world" empirical RD applications. The empirical exercises will employ R and Stata packages developed by the instructors. For further details, please visit the RD software website .

Prerequisites: Participants are assumed to have elementary working knowledge of statistics. It would be useful, but is not required, for participants to be familiar with recent developments in the literature on program evaluation and treatment effects. The course is designed to be self-contained and hence most underlying statistical and related concepts are introduced and explained along the course.

Fee: Members = $1500; Non-members = $2800

Tags: regression discontinuity, quasi-experimental methods, program evaluation, treatment effects

Course Sections

Section 1

Location: ICPSR -- Ann Arbor, MI

Date(s): May 31 - June 2

Time: 9:00 AM - 5:00 PM

Instructor(s):

  • Rocio Titiunik, University of Michigan
  • Sebastian Calonico, University of Miami

Syllabus: