Lesson 2 · Causal Inference · Step 1 of 6
Causal deniability
A study of coffee drinkers reports 5% lower mortality, states plainly it cannot establish cause and effect, and then suggests you drink coffee for your health. That move is worth naming, because it is everywhere. Demonstrating causality needs the cause to precede the effect, the two to be associated, and no plausible alternative to explain the association. Observational studies deliver the first two readily. The third is where confounding lives, and it is the one a disclaimer does not buy you.
Also covered: association and causation · three conditions for causality · confounding · internal validity
Alongside this step
- Slides and transcript — the whole video
- Read the chapter — Causal Inference
Lesson 2 · 6 steps
The Validity Toolkit · lesson 2 of 4