Several courses will be held on April 10. Each course up to a maximum of 60 participants.
For information on the fees and to register click here.
Workshop A: Causal Inference for Newcomers
Daniel Rhian (Cardiff University)
The workshop is aimed at newcomers to Causal Inference, and will cover the foundational topics needed to follow the majority of the conference talks.
Workshop B: DAGs for Causal Inference
Johannes Textor (Radboud University Medical Center)
DAGs can be useful to aid causal interpretation of observational data, but sensibly applying DAGs in practice can be very challenging. This workshop will cover (1) how we can test whether a DAG is mis-specified or inconsistent with a dataset; (2) how we can deal with the issue of statistical equivalence between DAGs; and (3) how we can perform DAG-based sensitivity analyses. The workshop will provide the necessary background theory but also focus on how to implement these methods using R and the R package `dagitty’.
Workshop C: Theory and Practice of Principal Stratification Analysis
Fabrizia Mealli (University of Florence)
The course will introduce the concept of Principal Stratification (PS), which was first formalized by Frangakis and Rubin (2002), but has its roots in the Instrumental Variables (IV) literature. But PS is a lot more than IV! PS has been applied to analyze causal effects in different settings, allowing to deal with various “selection” problems, such as censoring due to death, noncompliance, missing outcomes, mediation analysis, interference, and applied in various fields including economic, social and medical studies, using different modes of inference (moment- based, likelihood-based and Bayesian). The course will blend theory and applications.
Workshop D: Matrix Completion Methods for Causal Inference
Guido Imbens (Stanford Graduate School of Business)
This course will discuss recent methods for causal inference in panel settings. These methods link matching-type methods for estimating causal effects under unconfoundedness, synthetic control methods and difference-in-differences methods. These methods use techniques from the machine-learning literature on matrix completion and the econometric literature on interactive fixed effects. Prof. Imbens will discuss some empirical applications to illustrate the applicability of the methods and comparisons to previous methods.