What is expected when an independent random variable is added as a common cause during refutation?
StatisticsLevel: IntermediateCausal stress testing
What is expected when an independent random variable is added as a common cause during refutation?
Hints
- The added variable contains no causal information.
- Compare with the original estimate.
Answer
The estimate should remain reasonably stable within sampling variability. A large systematic change indicates an unstable estimation procedure.
Source and licence: DoWhy — random common cause test · MIT · answer: Quantoros editorial