Causality Case Studies for the Digital Causality Lab

About this repository

In this repo, we host a list of case studies used in the Digital Causality Lab (DCL).

The participants of the DCL develop data products (reports, blog posts, apps, packages etc.) in the context of these causality case studies. The content and idea of the case studies are described in issues of this repository.

The final data products will be shared in the DCL Gallery.

Scope of the case studies

Currently, we have five categories for the case studies.

1. Illustration of causal phenomena

• Illustration of typical biases, paradoxes and fundamental challenges to causal inference
• Examples: Omitted variable bias, collider bias / bad controls, $\ldots$

2. Graphical approaches

• Introduction and illustration of graphical approaches (DAGs)
• Examples: DAGs for collider bias, omitted variable bias, $\ldots$

3. Real-data examples

• Replication of empirical examples from research papers or textbooks

4. Introduction to causal estimation approaches

• Illustration of approaches to estimate causal effects
• Example: Instrumental variables, difference-in-differences, synthetic control, $\ldots$

5. Software for causal estimation in practice

• Implementation and demonstration of causal estimators
• Example: (Augmented) inverse probability weighting, g-formula, 2-stage least squared, $\ldots$

Add a new case study

To add your own idea for a case study, you can create a new issue. We prepared a template for you where you can describe the basic topic, ideas for the data product as well as links to references.

Contributing to the DCL

We host all source code of the DCL data products on GitHub (see the overview of our GitHub repositories. You can report bugs and improvements in the corresponding repositories.

In case you have questions or comments, feel free to contact Philipp Bach.