applying and researching causal inference

Causal Inference: Introduction

Getting started in causal inference is not easy as different scientific fields have different perspective on what causality means and how to quantify it. Here is a list of books that can help you get the idea of causal inference, what it's philosophy is and how to apply it. This list is based on this more extensive reading list.


Personally, if you are committed, I highly recommend Hernan's "Causal Inference Book". It's first 10 chapters teach you all the necessary basics, both potential outcomes and graphical models, of causal inference without requiring any real skill in Statistics.

  1. If you want a 3 chapter long brief introduction by Judea Pearl himself, you should read Pearl's Causal Inference in Statistics: A Primer (his words:

  2. If you want a quick blog post introduction, I recommend Ferenc's entry here.

  3. If you are not interested in the math at the moment, but want to get a more colorful read (popular science), you want to read Pearl's The Book of Why

  4. If you are looking for an overarching perspective that combines Potential Outcomes, Graphical Models and more, you should read Decision-theoretic foundations for statistical causality (paper, video)

  5. CODE: Here are some excellent scripts that discuss causal inference:


  1. If you want to get started with the research, you can read Peter's Elements of Causal Inference (free), but you might find it a bit too formal.

  2. If you want some great animations, I recommend Huntington-Klein's post (suggested by Jeremy Zucker)

  3. If you want to read a classic, I recommend CMU-based Causation, Prediction and Search.


  1. Pick more old and new papers from here.

  2. Send me an email if you are looking for something specific and I can help you find it.

  3. Find it on Google Scholar etc.

A map of causal inference

I created this map of causal inference to give beginner's an overview of where what kind of research is being done. Evidently, most of the research is driven by supervisor-student relationships e.g. students from MPI stay machine learning driven while students from Pearl behave similar to Pearl.