Part I Causal inference without model

1. causal effect and association

causation

association

2. randomized experiments

3. obsevational studies

4. effect modification

5. interaction

6. graphical representation of causal effects

7. confounding

8. selection bias

image-20220324232802016

9. measurement error

Part II causal inference with models

G-estimation

Doubly robust estimator

只要outcome model和treatment model 其中之一正确,doubly robust estimator 就是unbiased

  1. estimate WA=1f(A|L)
  2. fit E(Y|A=a,L=l,R), where R=WA if A=1, R=WA if A=0.
  3. standardize conditioning on L, then use E(Y|A=1,R)E(Y|A=0,R)E(Ya)E(Ya=0)

instrumental variable estimation

Part III Causal inference for longitudinal data

treatment-confounder feedback

image-20220331214028555

Conclusion