EasyDeepLearn

Potential outcomes framework — Rubin causal model.

hard

Answer

  • For each unit i and treatment T, define Yi(0)Y_{i}(0) (outcome if untreated) and Yi(1)Y_{i}(1) (outcome if treated).
  • Individual causal effect: Yi(1)    Yi(0)Y_{i}(1)\; - \;Y_{i}(0) — fundamentally unobservable (fundamental problem of causal inference).
  • Estimands: ATE = E[Y(1) - Y(0)], ATT  =  E[Y(1)    Y(0)    T=1]\mathrm{ATT}\; = \;E[Y(1)\; - \;Y(0)\; \mid \;T = 1].
  • Identification requires: (1) SUTVA (no interference / one version of treatment), (2) unconfoundedness (Y(0),  Y(1)    T    X)(Y(0), \;Y(1)\;\;T\; \mid \;X), (3) overlap.
  • Foundation of modern causal inference.
Check yourself — multiple choice
  • Random
  • Yi(0)Y_{i}(0), Yi(1)Y_{i}(1) potential outcomes; ATE = E[Y(1) - Y(0)]; identification needs SUTVA + unconfoundedness + overlap
  • Same as OLS
  • Not real

Rubin: Yi(0)Y_{i}(0), Yi(1)Y_{i}(1) potential outcomes; ATE / ATT under SUTVA + ignorability + overlap.

#causal-inference#theory

Practise Statistics Fundamentals

215 interview questions in this topic.

Related questions