Skip to content
Quantoros

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

Related questions