Seminar two

Question 1

What is Causal inference?

Question 2

Is there an attitude that causation cannot be determined with statistics?

Answer 2

Yes.

Question 3

Why is causation important?

Answer 3

Causation is important to consider, as it is these types of relationships that allow us to address problems, while it also allows to implement strategies that cause good outcomes.

Question 4

When thinking about causation using this formulation: p(O=1|do(E=1))-p(O=1|do(E=0)), what role does ‘do’ play in determining causation?

Question 5

What is the difference between confounder variables and mediator variables and collider variables?

Question 6

When designing experiments to attempt to determine causation, should you control for counfounding variables AND/OR mediating variables AND/OR colliding variables?

Answer 6

Control for confounding variables/factors.
DO NOT control for mediating variables/factors.
DO NOT control for collider variables.

Question 7

What is causal model theory discussed in seminar?

Question 8

Is making causal inferences based on correlational data hard?