Python can do causal inference too — introducing DoWhy

During my PhD, I believed advanced statistics belonged to specialized languages like R or Stata. I'm thinking of topics from the later chapters of Microeconometrics Using Stata by A. Colin Cameron and Pravin K. Trivedi — multinomial models (Chapter 18), evaluation of RCTs (Chapters 24–25), and the like.

Turns out Python does offer a package for causal identification: DoWhy. It models cause-effect relationships with explicit assumptions, separates identification from estimation, and can auto-validate its assumptions. It could solve my issues for those later chapters.

I still haven't taken the time to go deeper into causal inference, and DoWhy adds to the resources I want to try when I take that leap. Another relevant source is Causal Inference: The Mixtape by Scott Cunningham.