EPI 201 - Causal Inference

Question 1

Consistency

Answer 1

If Ai=a, then Yi^a = Yi^A=Yi. Therefore, Pr(Y^a=1|A=1) =Pr(Y=1|A=1)

Question 2

Conditional exchangeability

Answer 2

Y^a || A | L=l for all a

Question 3

What do you need for a question to not be too vague?

Answer 3

Clear definition of the outcome
Clear definition of the exposure including how it would be manipulated (e.g., not just “obesity”, but an understanding of how obesity would be intervened on – starvation, physical activity, etc)

Question 4

Characteristics of an ideal randomized trial

Answer 4

• Compliance to assigned treatment
• Double blind assignment
• (Near) Infinite population
• (No) Loss to follow-up
• Random treatment assignment
• Well-defined treatment
• Well-defined outcome

Question 5

Information missing in sufficient/component cause model

Answer 5

• dose of component causes
• preventive vs. harmful components
• temporal order/sequence of the component causes
• timing of action (induction periods) of the component causes