Why must causal identification specify which graph variables are actually observed?
StatisticsLevel: IntermediateCausal identification and targets
Why must causal identification specify which graph variables are actually observed?
Hints
- Availability constrains formulas.
- Distinguish latent and measured nodes.
Answer
An adjustment formula requiring unavailable variables cannot be evaluated from the dataset. A graph alone does not imply all its variables are measured.
Source and licence: DoWhy — identification before estimation · MIT · answer: Quantoros editorial