Teaching

Causal Inference with Directed Acyclic Graphs (MOOC)

This course is hosted on: Udemy.com

This course is an introduction to causal data science with directed acyclic graphs (DAG). DAGs combine mathematical graph theory with statistical probability concepts and offer a powerful approach to causal reasoning. Originally developed in the computer science and artificial intelligence field, they nowadays gain more and more traction also in other scientific disciplines (such as, e.g., machine learning, economics, finance, health sciences, and philosophy). DAGs allow to check the validity of causal statements based on intuitive graphical criteria, that do not require a any algebra. In addition, they open up the possibility to completely automatize the causal inference task with the help of special identification algorithms. As an encompassing framework for causal thinking, DAGs are becoming an essential tool for everyone interested in data science and machine learning.

The course provides a good overview of the theoretical advances that have been made in causal data science during the last thirty year. The focus lies on practical applications of the theory and students will be put into the position to apply causal data science methods in their own work. Hands-on examples, discussed in the statistical software package R, will guide through the presented material. There are no particular prerequisites for participating. However, a good working knowledge in probability and basic programming skills are a benefit.

Section 1: Introduction (Slides)
Section 2: Structural Causal Models, Interventions, and Graphs (Slides, R Code)
Section 3: Causal Discovery (Slides, R Code)
Section 4: Confounding Bias and Surrogate Experiments (Slides, R Code)
Section 5: Recovering from Selection Bias (Slides, R Code)
Section 6: Transportability of Causal Knowledge Across Domains (Slides, R Code)

Here’s the introduction video to the course on YouTube:

Causal Inference in Management and Economics Research

This two-day seminar provides an overview about methods for causal inference in empirical economic research. Starting with a general discussion about the concept of causality and its importance for addressing management and policy questions, it covers the most widely applied identification strategies in management, economics and computer science. Accompanying examples in Stata and R illustrate the discussed estimation techniques in a hands-on fashion. Due to its conceptual focus, the course is suitable for students with an intermediate level of knowledge in statistics and econometrics.

Lecture 1: The fundamental problem of causal inference (Slides)
Lecture 2: Dealing with confounding bias (Slides, Stata code)
Lecture 3: Identification with the help of surrogate experiments (Slides, Stata code)
Lecture 4: Regression discontinuity designs (Slides, Stata code)
Lecture 5: Using longitudinal data for causal inference (Slides, Stata code)
Lecture 6: Causal mediation analysis (Slides, R code)