Reviewing Consistencies

less than 1 minute read


In previous posts we have discussed abstractions between causal models; we have evaluated the abstraction error between a micromodel to a macromodel with respect to a property of interventional consistency.

In this notebook we review different forms of consistency (observational, interventional, and counterfactual), we analyze how we could evaluate them, and we discuss their meaning. Finally, we present a series of case study illustrating how different forms of consistency may or may not hold between SCMs related via an abstraction.

Notebook here