AIPW and double robustness

Combine outcome predictions with a propensity-weighted correction, then make either model—or both models—too simple.

What does double robustness guarantee?

AIPW can consistently estimate an identified effect when either the outcome model or the treatment model is correctly specified, even if the other is wrong. It combines predictions with weighted residuals rather than choosing one adjustment strategy.

Try it in the experiment

  1. Start with both models capturing the curved relationships.
  2. Make one model too simple, then restore it and change the other.
  3. Make both too simple and redraw several samples.

Double robustness is not protection against every causal failure. It does not fix unmeasured confounding, invalid adjustment, absent overlap, or two misspecified models. Nor does it promise that AIPW is closest to truth in every finite sample.