Read on Twitter
@Causalinf asked what my thoughts were on various online casual inference courses on the different platforms (for free) that I took. At least those were free when I took back in 2012-2018.
The very first one I took was the University of Copenhagen's Measuring Causal Effects in the Social Sciences ( https://www.coursera.org/learn/causal-effects). After this I was able to prove randomization leads selection bias to go to zero, then I could understand Mostly Harmless's logic.
Then I took MIT's Evaluating Social Programs ( https://www.edx.org/course/evaluating-social-programs-3), which was more applied than Copenhagens'. My primary learning was the concept of precision and reliability. Ester Duflo teaches one of the weeks.
@_MiguelHernan's Causal Diagrams: Draw Your Assumptions Before Your Conclusions in edX ( https://www.edx.org/course/causal-diagrams-draw-your-assumptions-before-your) was most enlightening. It opened the door to understanding @yudapearl's DAGs. This course is very easy to understand but deep in the thought process.
@PHuenermund's Causal Data Science with Directed Acyclic Graphs in Udemy ( https://www.udemy.com/course/causal-data-science/) is very satisfying as well. It mainly revolves around Pearl's causality. Take @_MiguelHernan's Causal Diagrams before this one.
I also love Duke University's causal inference course. It's understandable and interprets Card and Krueger's regression tables and other seminal papers. https://online.duke.edu/course/causal-inference-with-r-regression/
University of Pennsylvania's A Crash Course in Causality: Inferring Causal Effects from Observational Data ( https://www.coursera.org/learn/crash-course-in-causality) was short and sweet. My primary learning was the doubly robust estimator.
I skimmed (new course in 2018/19) the Columbia University's Causal Inference ( https://www.coursera.org/learn/causal-inference) and Causal Inference 2 ( https://www.coursera.org/learn/causal-inference-2). This course is more theoretical. Understanding big O and small o helps to conceptualize proofs.
