Class 6 - Spurious correlations
Bottom line: you need to make assumptions to identify causal effects, and those assumptions *are not testable with data*. Which means you have a model. You just either have an explicit model or one in your head, but you always need assumptions for identification. Always. Try DAGs
— scott cunningham (@causalinf) April 17, 2018
Zoom
Zoom: 12:00pm - 2:45pm, February 21, 2022
Required Readings
Chapter 5: The many variables & the spurious waffles
Chapter 6: The haunted DAG & the causal terror
Lectures
Lecture 5
Lecture 6
Deliverables
Due before class: Monday, February 21 at 11:59am
Lab for Class 6
Problem Set 4
Due by next class: Monday, February 28 at 11:59am