Scientific Coordination
Alisa Remizova
Administrative Coordination
Janina Götsche
Please wait...
Causal Inference with Instrumental Variables and Regression Discontinuity Designs
About
Location:
Online via Zoom
Online via Zoom
General Topics:
Course Level:
Format:
Software used:
Duration:
Language:
Fees:
Students: 220 €
Academics: 330 €
Commercial: 660 €
Keywords
Additional links
Lecturer(s): Martin Huber
Course description
This course will offer an in-depth exploration of two powerful methods for causal inference: Instrumental Variables (IV) and Regression Discontinuity (RD) designs. These techniques are particularly valuable for uncovering causal relationships in complex, real-world scenarios where experimental designs are impractical. The course will begin with an introduction to the core concepts underlying the methods and then go far beyond the basics, overall providing a comprehensive understanding of the theoretical foundations, advanced applications, and the latest methodological advancements.
IV methods are a popular tool in the arsenal of causal inference, providing a mechanism to uncover the causal impact of a treatment, such as participation in training, on an outcome of interest, such as earnings. Despite potential confounding factors or error terms that may influence both the treatment and the outcome, IV methods enable researchers to isolate the treatment effect. This course will be dedicated to the exploration of IV methods for nonparametric identification of the local average treatment effect (LATE). Beginning with an overview of the fundamental assumptions and concepts underlying instruments and the LATE, the course will progress to cover advanced topics, such as integrating machine learning algorithms, testing IV assumptions, and conducting sensitivity analyses.
RD designs are another powerful tool in the causal inference toolkit, for estimating causal effects in situations where treatment assignment (e.g., access to college education) is determined by crossing a threshold in a running variable (e.g., achieving a minimum test score). RD designs enable researchers to evaluate causal effects at the threshold, under the assumption that individuals just above and below the threshold are comparable in their background characteristics. This course will explore both sharp RD designs, where treatment assignment is strictly determined by the threshold, and fuzzy RD designs, where the threshold acts as an instrumental variable for the treatment. Starting with an introduction to the core concepts, the course will cover techniques for testing the RD assumptions, incorporating covariates, and leveraging machine learning.
Besides the econometric/statistical concepts of the various topics, the course will also discuss empirical applications from the social sciences and demonstrate the practical implementation of several methods in the statistical software package “R”.
Organizational Structure of the Course
The course will take place in the form of online lectures with group discussions of the topics, followed by lab sessions with data applications using the statistical software R. During the lab sessions, the lecturer will be available for individual support and consultations.
Target group
You will find the course useful if you are familiar with basic concepts of causal inference and would like to gain or deepen your knowledge about instrumental variable methods and/or regression discontinuity design and keep up with recent developments.
Learning objectives
By the end of the course, you will:
- understand the intuition, strengths, and potential shortcomings of various instrumental variable and regression discontinuity approaches,
- be able to apply instrumental variable methods and regression discontinuity designs to real-world data using the statistical software R.
Prerequisites
- Basic knowledge of regression analysis and causal inference (e.g., potential outcome notation or DAGs).
- Basic knowledge of the statistical software R. An introduction to “R”, which is free of charge, can be found here: http://cran.r-project.org/doc/contrib/Farnsworth-EconometricsInR.pdf
Software requirements
Statistical software R (https://cran.r-project.org/) and its interface R Studio (https://posit.co/) must be installed and ready to use on your PCs or laptops prior to the course
