Assumptions

In observational studies, conditioning on propensity scores or using g computation can lead to unbiased estimates of the exposure effect under:

1. Consistency
2. Exchangeability
3. Positivity